Guide
Marathon pacing strategy explained: fuel before fitness
The marathon is a fuel problem before it is a fitness problem. How pacing manages a finite store of carbohydrate, and why a fast first mile costs minutes.
By CricketTaken EditorialPublished Guide21 min read
Twenty-six miles is not a long way to travel. It is a long way to run at speed, which is a different problem, and the difference is chemical rather than muscular.
Marathon pacing strategy explained honestly starts in a store cupboard rather than at a stopwatch. The body runs on two fuels. One of them is effectively unlimited even in a very lean runner. The other is small, finite, and is the only one that can be burned fast enough to hold a decent marathon pace. The race is exactly long enough to empty it.
That is not an accident of the distance. It is the distance. A 10km race ends long before the cupboard is bare. A half marathon usually does too, which is why a half feels like a hard effort rather than a different kind of event. The marathon is the shortest common race at which the fuel supply, rather than the engine, decides the outcome for most of the field.
Everything that follows comes out of that one fact. Even splits, negative splits, hills, heat, gels, pace groups, heart rate straps and the horrible mile between twenty and twenty-one are all the same question asked in different ways: at what rate are you spending a store you cannot refill fast enough, and will it last?
Get that right and the race is difficult. Get it wrong by a small margin in the first hour and the race becomes a walk.
The marathon is a fuel problem before it is a fitness problem
Working muscle burns fat and carbohydrate. Both routes end in the same currency, the molecule that actually powers a contraction, but they are not interchangeable in the way that matters here.
Fat is abundant. Even an athlete carrying very little of it has more than any single race could consume, and nobody has ever finished a marathon having run out of fat. Fat is also slow. Getting energy out of it requires oxygen, a long chain of steps, and more oxygen per unit of energy released than carbohydrate needs. There is a ceiling on how fast a muscle can produce energy this way, and for almost everybody that ceiling sits below marathon pace.
Carbohydrate is the opposite. It is stored as glycogen, chained-up glucose held inside the muscles themselves and in the liver, plus a small amount of glucose circulating in the blood. It breaks down quickly. It yields more energy per litre of oxygen than fat does, which matters enormously when oxygen delivery is the thing being rationed. And it can be used without oxygen at all when demand outruns supply.
The catch is the size of the store. Muscle glycogen is measured in a couple of hours of hard running, not in days. Liver glycogen is smaller still, and its main job is keeping blood glucose steady for the brain rather than powering the legs.
So the marathon asks a question no shorter race asks. Can you reach the finish before the carbohydrate runs out?
One further wrinkle makes the answer less forgiving than it looks. Fat oxidation does not proceed independently of carbohydrate. The two pathways are linked, and when glycogen falls very low the machinery that burns fat runs less well too. The store does not simply run down and hand over cleanly to the other fuel. It runs down and takes some of the other fuel's capacity with it.
That is why the failure at the end of a badly paced marathon is so much more abrupt than a gradual fade. It is not a smooth handover between fuels. It is a cliff.
Why intensity, not distance, decides which fuel you burn
The mixture of fat and carbohydrate a runner is burning at any moment is not fixed. It slides with intensity, and it slides in one direction.
Jog slowly and most of the energy comes from fat. Speed up and the proportion coming from carbohydrate rises. Keep speeding up and carbohydrate takes over almost entirely. This is not a switch that flips at some particular pace, it is a continuous shift, and where an individual sits on it depends on their training, what they ate in the days before, the temperature, and how long they have already been running.
Three mechanisms drive the shift and they reinforce each other.
The first is oxygen economy. Carbohydrate yields more energy per litre of oxygen consumed. As pace rises, oxygen delivery becomes the binding constraint, and the body preferentially uses the fuel that gets more work out of the oxygen available. That is a sensible trade over ten minutes and an expensive one over three hours.
The second is speed of supply. Fat has to be mobilised from storage, carried in the blood, taken into the muscle cell and then into the mitochondria before any of it can be burned. Every one of those steps has a rate limit. Glycogen is already inside the muscle fibre, sitting next to the machinery that will consume it. When demand rises sharply, only one of these fuels can answer quickly.
The third is which fibres are doing the work. As pace increases, the muscle recruits fibre types that lean more heavily on carbohydrate and are less good at using fat. You are not simply asking the same muscle to work harder. You are bringing in different muscle, and the new recruits have different appetites.
The consequence is the most important thing about marathon pacing and it is rarely stated plainly. Pace is a fuel choice. Not a comfort choice, not an effort choice, a fuel choice. Deciding to run fifteen seconds a mile faster is deciding to draw down the small store faster, and the extra draw is not proportional to the extra speed. Because the fuel mixture is shifting at the same time as total energy demand is rising, a modest increase in pace can produce a disproportionately larger increase in the rate at which carbohydrate disappears.
Where those transitions sit for a given runner is the subject of the threshold testing that underpins most modern training prescription, and the published curves describing the relationship do not agree with one another about their placement or about how wide the individual variation is. Two runners with identical race times can burn noticeably different mixtures at the same pace. Training shifts the whole relationship, which is part of what a long marathon build is actually doing to you, and it is a large part of what separates a runner on their fifth marathon from an equally fit runner on their first.
What is not in dispute is the direction. Faster costs carbohydrate at more than a proportional rate. That asymmetry is why the arithmetic of pacing errors is so lopsided.
Marathon pacing strategy explained: one account, one withdrawal rate
Think of it as a single account with a fixed balance at the gun.
The balance is whatever glycogen you have stored, which depends on training, on the taper and on what you ate in the days beforehand. You cannot check it. There is no gauge, and the first reliable reading you get is the moment it runs out.
The withdrawal rate is set by your pace, adjusted for terrain, wind, temperature and how efficiently you happen to run. You control it, but only indirectly and only approximately.
Deposits are possible during the race, and they are small and slow, capped not by how much you can swallow but by how fast carbohydrate can cross the wall of your gut and reach the blood. That limit is real and it is not negotiable by effort.
Pacing is therefore the problem of choosing a withdrawal rate such that the balance reaches zero at the finish line and not one metre before it.
There are two ways to get it wrong. Withdraw too slowly and you finish with fuel unspent, which means you ran slower than you could have and you cannot retrieve those minutes. Withdraw too fast and you run out, at which point the pace collapses regardless of what you want.
These two errors are not the same size.
Finishing with a little unspent is a small loss, and it is a loss you can partly detect and correct while the race is still running. At twenty miles a runner who has been slightly conservative knows it, feels it, and can spend the surplus over the last six.
Running out is a large loss and it is uncorrectable. No manoeuvre available at twenty-two miles puts glycogen back into a leg. You can slow down, you can walk, you can take a gel and hope, and none of them restores the pace you had ten minutes earlier.
That asymmetry is the whole argument for conservative pacing, and it has nothing to do with timidity. It is the correct response to a penalty that punishes one direction of error far more severely than the other. A runner who aims a fraction under their true capability is not being cautious. They are being numerate.
- Miles 1 to 1080min
- Miles 11 to 2080min
- Miles 21 to the finish50min
Constructed illustration built on an invented runner targeting eight minute miles, which comes out a shade under three hours thirty for the full distance. Minutes, rounded. The point is the proportions, not the target.
Show the numbers
| Item | Value |
|---|---|
| Miles 1 to 10 | 80min |
| Miles 11 to 20 | 80min |
| Miles 21 to the finish | 50min |
The last block is under a quarter of the total time and carries nearly all of the risk. Every decision made in the first block is a decision about what will be available in the third.
The first three miles feel easy because they are the most dangerous
Every marathon opens with a lie, and the lie is told by your own body.
At the gun you are as fresh as you will ever be. Glycogen is full. You have tapered, so the legs are unusually willing. Adrenaline is up, the crowd is loud, the temperature is at its lowest point of the morning, several thousand people around you are running faster than they should, and many big city courses start downhill because that is where the wide roads happen to be.
Under those conditions, target pace feels like a jog.
This is not an illusion in any interesting sense. Perceived effort is a real signal, but it does not measure pace. It measures pace plus the accumulated cost of everything already done, and at mile two the second term is close to zero. The same sensation of effort corresponds to a much faster pace at the start of a marathon than it does at mile twenty, and the gap between those two paces is the whole event.
So the error is invisible while you are making it. Ten seconds a mile too fast does not feel like ten seconds a mile too fast. It feels like today is a good day.
Several practical things push in the same direction. Crowding in the opening mile makes runners weave and surge to find room, and weaving adds distance. Tall buildings degrade the satellite signal a watch relies on, so the pace readout in the first miles of a city marathon is frequently wrong. Courses are measured along the shortest legal route, so a runner who does not take the tangents through bends covers more ground than the distance markers claim, and their true pace is faster than their watch pace by more than they realise.
Now the arithmetic. Take an invented runner aiming at eight minute miles, which comes out a shade under three hours thirty. Suppose that on the morning it feels easy and they drift twelve seconds a mile quicker for the first five miles, ten seconds quicker for the next five, and four seconds quicker for the five after that, before the bill arrives.
The gains total a hundred and thirty seconds. The losses total four hundred and thirty-two. This invented runner finishes about five minutes slower than an even pace would have delivered, having spent the first hour feeling comfortable and ahead of schedule.
That is the shape of the thing. Small gains banked early at full price, large losses paid late with interest. The opening miles feel like the cheapest part of the race and they are the most expensive, because the currency they are spent in is the one that runs out.
A second-order effect makes it worse. A runner who is ahead of schedule at halfway rarely responds by slowing down. They respond by defending the buffer, which means holding the fast pace for longer, which means the eventual collapse starts from further down the fuel gauge. The buffer is not insurance. It is a debt with a repayment schedule you do not get to choose.
The wall is a fuel gauge reading, not a character test
Wall is a bad metaphor for what happens, and the metaphor has done real damage to how runners think about it.
A wall suggests a barrier: something in front of you that you either break through or fail to break through. That framing turns a metabolic event into a moral one, and it produces exactly the wrong response, which is to grit the teeth and push for one more mile before conceding anything.
What actually happens is a ceiling coming down. As muscle glycogen falls, the rate at which the muscle can generate energy falls with it, because the fast fuel is no longer available in the quantity the pace demands. What remains is fat oxidation, plus whatever glucose is in the blood, plus whatever has been absorbed from a drink or a gel in the previous twenty minutes. That combination supports a real running speed. It does not support your marathon pace.
So the pace drops. Not because you decided to slow, and not because it hurts more, but because the supply will not fund the demand. Runners describe the sensation accurately when they say the legs stopped answering. Effort goes up, output goes down, and the gap between the two widens with every mile.
A separate and simultaneous event gets lumped in under the same word. Liver glycogen keeps blood glucose stable, and the brain runs almost exclusively on blood glucose. When liver glycogen empties, blood glucose falls, and the brain notices before the legs do. The symptoms are not muscular at all: a flat grey misery, difficulty with simple decisions, an urgent conviction that stopping would be entirely reasonable, sometimes a strange loss of interest in the race. Runners who have had it describe the moment they stopped caring. That is a chemical state, not a personality flaw, and it is one reason carbohydrate taken during a race helps even in quantities too small to make much difference to the muscles.
Several other things produce a late-race collapse that looks identical from the roadside, and telling them apart matters because the responses differ.
Muscle damage from repeated impact, particularly on a course with sustained downhills, degrades the muscle's ability to produce force regardless of how much fuel is available. Cramp is a different failure again, still not properly explained, and not reliably fixed by any of the things people try mid-race. Dehydration and heat strain reduce the blood volume available to muscle and skin at the same time. Gastrointestinal shutdown stops you absorbing the fuel you are taking, producing a fuel failure by a different route and adding nausea to it.
A runner who slows at twenty-two miles is often experiencing two or three of these at once. The single-cause story is tidy and usually wrong.
What is reliably true is that most late-race fuel failure traces back to a decision made in the first hour, when everything felt fine. The wall is not where the mistake happens. It is where the mistake becomes visible.
Even splits, negative splits and what the evidence actually supports
Three descriptions of a race are in common use. A positive split means the second half was slower than the first. A negative split means it was faster. An even split means the two halves matched.
Look at the finishing data from any large marathon and positive splits dominate. That much is not controversial. What follows from it is.
The pattern that holds most consistently across the research is about variation rather than direction. The better the runner, the smaller the fluctuation in speed across the race, and elite fields cluster far closer to even than the mass field does. Recreational runners slow considerably, and the slower the finishing time the larger the slowdown tends to be. Reviews of the literature also report a difference by sex, with women on average starting more conservatively and holding speed more steadily than men over the same distance.
Be careful about what that shows. Even splitting is at least as much a marker of accurate self-assessment as a cause of a good result. Runners who finish with matched halves are, by and large, the runners who picked the right target. Instructing a runner with an unrealistic goal to run even splits does not produce an even race, it produces the same collapse a mile later.
The research is also messier than the tidy three-way classification suggests. Studies define the categories differently, use different numbers of segments, and disagree about whether these are genuinely distinct strategies or simply different magnitudes of the same underlying pattern. A review that reads the same data through a coarser classification will find far more even pacing than one that slices the race into many segments. Terminology is inconsistent enough across the literature that comparing two studies directly is often not valid. Anyone quoting a single confident number about optimal splits is quoting past a real disagreement.
What the mechanism supports is narrower and more useful than the debate.
Running the first half faster than the second is expensive for the reasons already set out: the extra fuel is spent at a premium and never returned. Running the first half much slower than the second is also expensive, because a runner who finishes with a genuine surplus has been carrying fuel they could have converted into speed, and the conversion is no longer available once you cross the line.
The optimum sits close to even, with the balance of risk tilted slightly towards the conservative side. That is not a compromise, it is a direct consequence of the asymmetric penalty. Plan even. Run the first ten miles a shade under. Arrive at twenty miles with something to decide.
- Even plan
- Fast start
Constructed illustration using the invented eight minute mile runner from the previous figure. The even plan holds 480 seconds per mile throughout. The fast start plan opens quicker, holds it too long and finishes at nine minute pace. Seconds per mile. Neither line is measured from anything.
Show the numbers
| Item | Even plan | Fast start |
|---|---|---|
| Miles 1 to 5 | 480s | 468s |
| Miles 6 to 10 | 480s | 470s |
| Miles 11 to 15 | 480s | 476s |
| Miles 16 to 20 | 480s | 492s |
| Miles 21 to the finish | 480s | 540s |
The two lines cross between fifteen and twenty miles, and that crossing point is where most marathons are decided. Before it, the fast start looks like the better race. After it, the gap opens faster than any amount of determination can close.
The course is part of the plan, and an uphill start is the easy version
An even split by the clock is not an even effort, and on anything other than a pancake-flat course the two pull against each other.
Running uphill at target pace costs more than running flat at target pace, and the extra cost is disproportionate. Holding the clock steady up a climb means spending fuel at a rate the flat sections never demanded, which is why the standard advice to run hills by effort rather than by pace is sound rather than merely comforting. You give away seconds on the climb and take back most of them on the descent, and the fuel account stays roughly where the plan wanted it.
Downhill is where the reasoning gets more interesting, because a descent is cheap in one currency and expensive in another.
Running downhill costs less energy, so the fuel account barely notices it. What it does instead is load the muscle in a lengthening contraction, over and over, as the quadriceps work to control the descent rather than to produce it. That kind of loading causes more mechanical damage than the equivalent work uphill, and the damage does not announce itself while it is happening. It arrives later as a loss of force production, which looks and feels like a fuel problem but is not one, and no amount of carbohydrate fixes it.
That inverts the usual worry. A course with a hard uphill start is a nuisance and it is honest: it tells you at once what it is taking, and it forces the conservative early pace that a marathon wants anyway. A course with a long downhill opening is the dangerous one, because it hands out free speed at a discount that has to be repaid in the last hour, and it does so during exactly the miles when a runner is least equipped to notice.
A downhill finish is a different proposition again, and mostly a good one. Descending in the last few miles asks for force production from legs that are already damaged, so the free speed on offer is less free than the profile suggests, but it arrives at a point where there is no later to pay it back in.
Governing bodies take the shape of a course seriously enough to write rules about it. For a road performance to be eligible for record purposes, World Athletics limits the net elevation drop from start to finish, expressed per kilometre of race distance, and limits the separation between the start and the finish as a proportion of the race distance. The first stops a course being fast simply because it runs downhill. The second stops a course being fast simply because a prevailing wind pushes the whole field one way. The specific limits are published in the rules and are amended from time to time, which is why some genuinely quick courses on the calendar are not record eligible and say so plainly in their own entry material. Anyone comparing performances across the big city races that make up the marathon majors is comparing courses as much as runners.
Two other course facts are worth knowing before setting a pace target.
The measured distance is the shortest legal route, established by a calibrated bicycle along the line a runner could theoretically take. Nobody runs that line. Every metre you drift wide of it is a metre you have added to your own race at no benefit, and in a crowded field that adds up faster than most runners expect.
And the drinks stations are on the course whether you use them well or badly. A station in the middle of a narrow section, taken at full pace, produces a surge, a swerve and a spilt cup. Slowing marginally to take a drink properly costs a handful of seconds and returns more than that, which is the rare pacing decision where the conservative option is also the fast one.
Heat and humidity change the target, they do not test your resolve
Most of the energy a running muscle uses does not become forward motion. It becomes heat, and the body has to get rid of it or core temperature climbs.
It gets rid of it in two ways that both compete with running. Blood is sent to the skin, where heat can leave, and that blood is not available to the working muscle. Sweat is produced and evaporates, which is the only route that does meaningful cooling at running speeds, and producing it costs plasma volume from the blood.
Both effects show up as the same thing on a watch: heart rate rising at a constant pace as the race goes on. That upward drift is not a sign of poor fitness. It is a smaller volume of blood being pumped more often to do the same job while also cooling you.
Humidity matters more than air temperature and gets discussed less. Sweat only cools when it evaporates, and evaporation depends on how much moisture the surrounding air can still absorb. In humid air, sweat runs off the skin without taking much heat with it, which means the fluid cost is paid and the cooling benefit is not. Dew point is a far better predictor of how a marathon will feel than the number on the thermometer.
Now connect it back to the fuel account, because this is the part that gets left out.
Heat does not merely make a given pace less pleasant. It raises the proportion of energy coming from carbohydrate at that pace. A hot day therefore attacks the fuel budget directly, at the same time as it reduces how much you can absorb from the gut, and at the same time as it raises the cardiovascular cost of everything. Three separate mechanisms, all pushing the same way.
Which is why a target set in comfortable conditions cannot simply be carried into a hot race and defended. It has to be revised, and revised before the gun rather than at the point where it fails.
A revision made at the start is a slower finish. A revision forced at twenty miles is a walk. The two are not remotely equivalent, and the runner who insists on holding the original number until it breaks has chosen the second one without noticing that they were choosing.
Practically, this means deciding the adjustment during the warm up, when the information is available and the adrenaline has not yet arrived. Look at the dew point rather than the temperature. Look at how much of the course is exposed and at what time the sun reaches it. Note that a cool start with a rising sun is a different race from a warm start under cloud, even if the average temperature matches. Then use the first five kilometres as data rather than as a statement of intent: if the revised pace is already costing more than it should, revise again.
Cold water on the skin and ice held at the neck and wrists help more than they have any right to, mostly by changing what the brain thinks about the heat rather than by moving much energy. That is a legitimate benefit, not a trick, because perceived effort is one of the things regulating your pace whether you like it or not.
Fuelling during the race, and the limit nobody can train away
If the store is finite, the obvious move is to top it up as you go. That works, within a constraint that is harder than most runners assume.
The constraint is not the stomach. Swallowing is easy, and a runner can put down far more carbohydrate than they can use. The constraint is the intestinal wall, where carbohydrate has to be actively transported into the blood before any muscle can touch it, and where the transporters saturate.
Glucose, and the maltodextrin chains that break down into it, are absorbed largely by one transport route. Saturate it and additional glucose sits in the gut drawing water in, which is where the sloshing, the bloating and the emergency portable toilet come from. Fructose uses a different transporter. Because the two routes are separate, a mixture of glucose and fructose can be absorbed faster than an equal quantity of glucose alone, which is why almost every modern sports drink and gel is formulated as a blend rather than as a single sugar. That is the mechanism behind the ratios printed on the packaging.
The published guidance on how much carbohydrate per hour a runner should target is set out by sports nutrition bodies and has been revised upwards over the years as the mixed-transporter work matured. It is also individual, and the sensible number for a given runner is the one they have actually rehearsed rather than the one on the wrapper.
Two further limits sit on top of that.
Blood flow is one. Running sends blood to muscle and skin, and the gut receives less of it than at rest. Absorption falls as intensity rises and falls further in the heat, which means the ability to take on fuel is lowest at precisely the point in the race where the need is greatest. A plan that works comfortably at long run pace on a cool morning is not automatically a plan that works at race pace in the sun.
Time is the other. Carbohydrate taken now is not available now. It has to be swallowed, emptied from the stomach, absorbed, and delivered, and the whole sequence takes long enough that a gel at twenty-two miles does very little for mile twenty-three. Fuelling is a rate problem to be managed from early in the race, not a rescue to be deployed when the gauge reads empty. By the time a runner feels they need one, the useful moment to have taken it was half an hour earlier.
The gut is trainable, and this is one of the more solid practical findings in the area. Taking carbohydrate regularly during long training runs increases how much can be tolerated and absorbed later. It is a genuine adaptation, and it is one of the few marathon-specific things that can be improved in the final weeks of a build without adding fatigue.
Fluid follows a different logic and deserves its own sentence: the goal is to limit losses, not to eliminate them, and drinking far beyond thirst carries a real risk of diluting blood sodium, which is dangerous in a way that mild dehydration is not. Both extremes have hurt runners. Drinking to a rehearsed plan, adjusted for the conditions and for how much you actually sweat, is the position that avoids both.
One rule covers all of it. Whatever the plan is, it has to be the plan you practised, using the products the race is actually handing out at the tables. The drinks on course are whatever the organiser has contracted for, which is published in advance, and race morning is a poor time to discover that a particular formulation disagrees with you.
Drafting, pacers and why a record attempt is not a championship race
Air resistance at marathon speeds is a modest fraction of the total cost of running and it is not nothing, and it rises with the square of the relative air speed, which is why a headwind is punishing out of all proportion to how strong it feels. Running behind someone reduces it. Running in a group reduces it more.
The aerodynamic saving is real, and cycling has studied the effect far more rigorously than running has because the speeds involved make it dominant rather than marginal, as the work on what actually happens inside a bunch sets out. Running gains less from shelter than cycling does, and it gains something.
The larger benefit of running with others is cognitive, and it is badly underrated. Holding an exact pace for three hours is a continuous act of attention, and attention is a resource that depletes. A runner following someone else has outsourced the pace decision and only has to hold position, which is a much easier task. That is a substantial part of what a pacemaker provides, and it is why runners who lose contact with a group often slow by more than the aerodynamic arithmetic can explain.
The competition rules draw lines around all of this. Pacemaking within a race, by an athlete who has entered the race and is running it, is permitted in road racing and is a normal feature of time-focused events. What is not permitted is assistance from outside the race: pacing by anyone not competing, by a vehicle or a bicycle, or by a moving light or device. The distinction is between an entered competitor who happens to be running at a helpful speed in front of you and an external aid delivered to you during the event, and the exact provisions sit in the World Athletics competition rules alongside the conditions for record ratification.
That distinction explains a lot about why certain performances are ratified and others are not, and it explains why the fastest organised time trials over the distance sit outside the record book by design rather than by accident.
A championship marathon is a different sport played on the same course.
There are no pacemakers. Nobody is being paid to make the race fast. The medal is the entire objective and the clock is decoration, which changes the optimisation problem completely. A field that would each run their fastest by holding an even pace instead runs slowly, together, watching each other, because the runner who leads is spending fuel to shelter everybody behind them for no reward.
Then somebody surges, and the surge is a weapon precisely because of the fuel arithmetic set out earlier. Covering a surge costs the responder disproportionately, since the pace increase is short but the fuel drawn is drawn at the premium rate. The athlete who chooses the moment has decided they can afford it. Everybody who responds is paying a price they did not choose, at a moment chosen by someone else. Repeat that three or four times and the pack breaks, not because anyone was outsprinted but because the responders ran out of the ability to respond.
This is why championship marathons produce slow times and thrilling races, and why an athlete who races well tactically and an athlete who runs fast times are often not the same athlete. The ranking systems that decide who gets into those championships have to reconcile the two, which is harder than it sounds and is one of the reasons those systems are complicated.
For an ordinary runner in a mass race, the equivalent question is whether to run with an official pace group. The honest answer is that they help some people and hurt others. They provide the cognitive relief and some shelter. They are also crowded, they take the tangents badly because a group cannot, they compress and surge through drinks stations, and they are pacing to a clock time that may or may not be the right target for you on the day. A pace group is a tool, not a plan, and the runner who cannot leave one when the day turns out to be hotter than forecast has handed their own decision to a stranger with a flag.
Heart rate, pace and effort are three imperfect controls
Something has to regulate the pace once the gun goes. There are three candidates in common use and each of them is wrong in a different way.
Pace is objective, precise and completely blind to whether you can afford it. A watch showing target pace tells you nothing about the state of the account funding it. It is also less reliable than it appears: satellite positioning degrades badly among tall buildings and under bridges, per-mile splits taken from a watch drift out of alignment with the course markers as the accumulated distance error grows, and instantaneous pace readouts are smoothed guesses rather than measurements. Pace works best as a ceiling in the first half. Not a target to hit, a number not to beat.
Heart rate measures the response, not the output. It tells you what the effort is costing your cardiovascular system, which is genuinely useful information, and it is contaminated by almost everything: caffeine, adrenaline at the start, air temperature, sleep, a virus you have not noticed yet, how recently you drank. It also drifts upwards through a marathon at constant pace as plasma volume falls and heat load rises, so the same number means different things at mile four and mile twenty. And it lags. A change in effort takes tens of seconds to show up, which makes it useless for moment-to-moment control on undulating ground.
Where heart rate earns its place is as an early warning. If, at five kilometres, the heart rate at target pace is clearly above where the same pace has been sitting in training, the day is not the day. That is information worth having at five kilometres and worthless at thirty.
Perceived effort integrates everything. It is the only one of the three that knows about the fuel, the heat, the hills, the damage and the state of your morning simultaneously, and it is the control that experienced marathon runners actually use for most of the race. Its weakness is the one already described: it systematically understates the cost of the pace in the first hour, which is exactly when the consequential decisions are made.
The practical arrangement uses all three in sequence. Pace as a hard ceiling early, precisely because effort cannot be trusted then. Heart rate as a sanity check in the first few miles and then largely ignored. Effort as the governor through the middle of the race, when it has become honest. And in the last ten kilometres, effort is the only control still functioning, because pace has become an output rather than an input.
Running power meters have arrived as a fourth candidate, and the sensible position is scepticism rather than dismissal. Unlike cycling, where power is a directly measured mechanical quantity, running power is a modelled estimate, and different manufacturers model it differently enough that their numbers are not comparable. The idea is sound and the implementations disagree. Anyone using one should treat it as a well-behaved version of pace that accounts for gradient, which is genuinely useful on a hilly course, rather than as a direct measure of what the body is spending.
What a realistic marathon target actually rests on
A marathon target is not a wish, a round number or the name of the training plan you followed. It rests on three things, and only the first two are available before race week.
The first is recent evidence at a shorter distance. A recent half marathon or 10km time tells you about the engine: the oxygen delivery and the pace you can sustain before the metabolic cost runs away from you. Race-equivalence tables convert those times into a marathon prediction, and they are useful as long as their assumption is understood. They are calibrated on runners who have done marathon-specific training. Give the same table to a runner with good speed and modest weekly mileage and it will overpredict, sometimes dramatically, because a fast half proves you have the aerobic capacity to support the pace and proves nothing at all about whether you have the tank to hold it for twice as long.
That is the whole point of the fuel framing. Shorter races are limited by oxygen delivery and by tolerance of the metabolic by-products of hard running. The marathon is limited by fuel and by durability. They are different constraints, and being good at one does not settle the other.
The second is direct evidence of durability. The relevant question is not whether you can run the distance but whether you can hold something close to target pace late in a long run on legs that are already tired, without the pace falling away by itself. A build that includes sustained work at goal pace towards the end of long runs generates that evidence. A build that consists of easy long runs and short fast sessions does not, and a runner who has never held goal pace when tired is guessing.
The third is the conditions, which arrive on the day and subtract from whatever the first two supported.
Out of that comes the only planning tool that survives contact with an actual race: three targets, decided in advance. An A target that requires the day to go well, a B target that is realistic if it goes normally, and a C target that is a good outcome if the weather is hostile or the stomach rebels. Write them down before travelling.
The reason for three is not motivational. It is that the decision at twenty miles is enormously easier if it is a selection between prepared options than if it has to be invented on the spot by a person with low blood glucose. Reducing a hard judgement to a choice from a short list is the single most effective thing a runner can do for their own late-race decision making, and it costs nothing.
Equipment belongs in the same conversation and mostly at the margins. Shoes affect the energy cost of running and therefore the rate at which the account drains, which is a real effect on the fuel arithmetic rather than a marketing one, though the size of it for any individual is far less settled than the advertising suggests. It does not change any of the reasoning above. It changes the pace at which that reasoning is applied.
- Before the gunThe target is chosen from recent race evidence, from proof you can hold goal pace when tired, and from the forecast. Three targets are written down. This is where most marathons are actually decided.
- The warm upThe conditions are read and the target is revised if the day demands it. Dew point matters more than temperature. A revision made here costs minutes; the same revision forced at twenty miles costs far more.
- Miles 1 to 3Everything feels easy because fatigue has not accumulated yet, so effort understates pace. Crowding, weaving and poor satellite reception all hide the error. Pace is used as a ceiling here precisely because feel cannot be trusted.
- Miles 4 to 6The first honest data arrives. If the effort at target pace is higher than it has been in training, or heart rate sits above where that pace normally puts it, the target is wrong and the cheapest moment to change it is now.
- Miles 7 to 13The settled phase. Fuelling runs to a rate rather than to a feeling, because absorption takes time and cannot be caught up later. Hills are run by effort, not by pace, and time given away on a climb is taken back on the descent.
- HalfwayThe temptation to bank time is at its peak and is the most expensive thought available. A runner ahead of schedule here should let the schedule catch up rather than defend the buffer.
- Miles 14 to 18The fuel account is now visibly draining. Perceived effort begins telling the truth. A pace that required no attention at ten miles now requires some, and that change is itself the reading.
- Miles 19 to 20The last point at which a small correction still works. Easing marginally here preserves the ability to run to the finish. Holding on here is the decision that produces the walk two miles later.
- Miles 21 to 24Whatever was banked or spared is being spent. Pace has become an output rather than an input. Fuel taken now arrives too late to change much, so the only lever left is how hard the remaining fuel is drawn.
- Miles 25 to the finishDamage, not fuel, is often the binding constraint by this point, and it does not respond to anything except a lower pace. The gap between running slowly and walking is enormous, so the priority is to keep running at all.
A qualitative account of the sequence rather than a schedule. The mile markers are the conventional ones for the distance and the descriptions are mechanisms, not measurements.
The decisions at twenty miles
Twenty miles is where a marathon stops being a plan and becomes a set of choices, and it is worth knowing in advance what is actually being chosen.
By this point you have information you did not have at the start. You know what the pace has genuinely cost on this particular day, in this weather, on these legs. You know whether your stomach accepted the plan. And you have roughly ten kilometres left, which is short enough to estimate and long enough to ruin.
Three questions settle it, and they should be asked in this order.
Is the pace costing what it was costing ten miles ago? Not whether it hurts, because it will. Whether the relationship between effort and pace has shifted. If effort has climbed while the pace has stayed flat, that is the fuel gauge reporting, and it will not improve on its own.
Have you taken what you planned to take? If the fuelling plan slipped, either because the stomach rebelled or because a station was missed in the crowd, then the remaining distance is being run on a smaller balance than the plan assumed, and the plan is now optimistic by an amount you can estimate.
What is actually left worth? Not what you wanted before the start. What the remaining fuel and the remaining leg function will fund.
Three responses follow, and their consequences are not symmetric.
Holding pace is the right answer when effort and pace are still in the relationship they had at halfway. It is the wrong answer, and by far the most common error, when they are not. Holding a pace the account cannot fund does not preserve the time. It converts a slow finish into a stopped one.
Easing slightly, by a handful of seconds a mile, is the decision that pays most often and gets made least. It feels like surrender and it is arithmetic. A small reduction in pace cuts the withdrawal rate disproportionately, for the same reason a small increase raised it disproportionately, and it frequently buys enough to run the whole way to the finish rather than most of it.
Pushing is correct in exactly one situation: when the first half was genuinely conservative, the effort at target pace is still where it was at halfway, and the fuelling plan went in as intended. That is what all the earlier caution was for. A runner who arrives at twenty miles with something in hand has bought themselves the only good option in the race, which is the ability to choose to go faster.
The reason the easing decision matters so much is that the difference between running and walking is a cliff, not a slope. A runner who slows to a jog is still covering ground at a respectable rate. A runner who stops rarely resumes, and if they do it is at a pace they would have found insulting an hour earlier. Everything after twenty miles should be evaluated against that single threshold: does this decision keep me running?
Which returns to where this started. The marathon is a fuel problem before it is a fitness problem. The fitness sets the size of the account and the ceiling on what it can fund. The pacing decides whether the account is spent evenly across the distance or handed over in the first hour to a version of yourself who was feeling optimistic and had no idea what he was signing for. The rest of our athletics coverage works from the same view of the event.
Three numbers are worth knowing before the gun, and none of them is your goal time. The pace you have actually held for a sustained effort on tired legs. The rate of carbohydrate per hour your gut has handled in training without complaint. And the pace you would settle for if the dew point is high and the day is against you.
A runner who knows those three arrives at twenty miles with a decision to make. A runner who knows only a goal time arrives with nothing to decide, because the race has already decided it.
Common questions
How should I pace a marathon?
Choose the fastest pace you are confident you can hold for the whole distance, then run the first half at or just under it rather than above it. Running slightly too fast early costs far more time later than it gains at the time, because the fuel spent early is not recoverable, while running slightly too slow early is correctable at twenty miles. Plan an even pace, execute the first ten miles a shade conservatively, and let the last ten kilometres take whatever is left.
What is the wall in a marathon?
It is the point at which the muscles and liver run low on stored carbohydrate, so the rate at which the body can supply fuel falls below the rate the chosen pace demands. Pace then drops sharply even though effort rises, because fat oxidation cannot support marathon speed on its own. It is a metabolic ceiling coming down rather than a failure of will, and it is usually the result of a pacing decision made in the first hour.
Is a negative split better than an even split in a marathon?
The evidence supports running close to even, and fast marathons are generally run that way rather than with a large negative split. A slightly negative split is a sound practical target because it means you left a margin early that you could still spend late, and unspent fuel at the finish cannot be converted back into time. What the research is clear about is that large positive splits, where the second half is much slower, are the common failure among recreational runners and the signature of a target set too high.
How much should I slow down in the heat for a marathon?
Enough that the pace still feels like the effort you rehearsed, which almost always means revising the target before the gun rather than defending it until it breaks. Heat and humidity raise the cost of the same pace, increase carbohydrate use and reduce how much fluid and fuel your gut can absorb, so the day changes the fuel arithmetic and not just the comfort. Decide a revised target during the warm up, use the first five kilometres to test it, and accept a slower finish rather than a walk.
How often should I take a gel or drink during a marathon?
Often enough to keep a steady rate of carbohydrate arriving from early in the race, because absorption is limited by how fast carbohydrate crosses the gut wall and cannot be hurried once you are already low. Products combining glucose and fructose are taken up by two separate transport routes, which raises the ceiling compared with glucose alone. Practise the exact plan and the exact products in training, since the gut adapts to what it is regularly asked to handle and race day is the wrong place to find out.
Filed under Athletics·athletics · running · marathon · endurance · physiology