Lactate threshold explained, and why lactate is not the enemy

Somewhere in the second half of a hard session, a coach tells you that you are producing lactic acid, that the acid is what is burning your legs, and that tomorrow's stiffness is what is left of it. Three claims, delivered as one sentence. All three are wrong, and the last of them is not even in the right week.

Lactate threshold explained honestly has to begin by clearing that away, because almost everything people believe about the subject is downstream of it. There is no lactic acid in a working muscle in any quantity worth discussing. Lactate is not what makes your legs hurt. It has nothing whatsoever to do with the soreness that shows up two days later. It is a fuel, and your heart is burning some as you read this.

What the threshold marks is not the arrival of a poison. It is the intensity at which lactate starts entering the blood faster than the body takes it back out. That is a completely different proposition, and once the difference is clear, most of the confusion about training zones goes with it.

• 4Millimoles per litre in the most-used fixed threshold convention
• 2Thresholds in the standard two-threshold model
• 3Intensity domains the two thresholds create
• 1Rise in mmol/L that ends a maximal lactate steady state trial

These are definitional values used by convention in the literature, not measurements of any athlete.

The chemistry the story gets backwards

Lactic acid exists. It is just not what is in you.

An acid gives up a proton in water when the surrounding pH is above its dissociation constant. Lactic acid has a pKa a shade below 4. The inside of a muscle cell, even a badly stressed one, does not get remotely that acidic. So the moment lactic acid forms in a human being, it hands over its proton and becomes the lactate anion. What circulates, what the analyser reads, and what the whole industry is named after, is lactate. The acid is a laboratory abstraction here.

Now the part that inverts the folk explanation. Glycolysis breaks glucose down to pyruvate and releases energy. The protons that acidify a hard-working muscle come mostly from the other side of the ledger: every time a molecule of ATP is split to release its energy, a proton is liberated, and every time the mitochondria rebuild that ATP, a proton is consumed. At an easy intensity the two rates match and nothing accumulates. Drive the intensity up and ATP is being split faster than the mitochondria are rebuilding it, the accounting no longer balances, and free protons pile up.

Reducing pyruvate to lactate does the opposite. That reaction takes a proton out of the cell's pool. Producing lactate is, on the chemistry, mildly protective against the acidification it gets blamed for.

So why did the blame stick for the best part of a century? Because the two things move together. Lactate and protons both accumulate when glycolytic flux runs ahead of oxidative capacity, so blood lactate is an excellent marker of an acidifying muscle. It just is not the cause of it. Early muscle physiology worked on isolated frog muscle stimulated without oxygen, saw lactate accumulate as tension fell, and drew the obvious inference. It was the wrong inference from a good experiment, which is the most durable kind of error there is.

The burn itself is not one substance either. Hydrogen ions contribute. So does inorganic phosphate, released as ATP is split, which interferes directly with the machinery of contraction. So does potassium leaking out of the fibres. The sensation reaches you because small nerve endings inside the muscle respond to that metabolite mixture and report it upward. Lactate is one ingredient in the broth and, taken alone, a fairly innocuous one.

Lactate is a fuel, and most of it never leaves the traffic system

Here is the fact that reorganises everything else. Lactate is produced continuously, at rest, in tissue that has all the oxygen it could want. It is not an emergency product. It is a normal intermediate of carbohydrate metabolism, made every minute of your life.

The reason is structural. Fast-twitch fibres are built for glycolytic flux and carry the enzyme profile to match, so they turn glucose into pyruvate faster than their mitochondria can accept it. The overflow becomes lactate. Slow-twitch fibres, sitting in the same muscle, are the opposite: dense with mitochondria, comparatively poor at generating glycolytic flux, and very good at oxidising anything that arrives. The lactate leaves one and enters the other.

That transfer is not passive diffusion. It runs through a family of transporter proteins in the cell membrane. One of them predominates in glycolytic fibres and handles export. Another predominates in oxidative fibres and handles import, and it also sits on the mitochondrial membranes inside those fibres. The traffic is deliberate and it is regulated, which is not how a body treats rubbish.

The destinations go well beyond the neighbouring fibre. The heart takes up lactate from the blood and burns it in preference to glucose when the supply is good, which is convenient, because the heart is working hardest exactly when the supply is highest. The brain uses it. The liver takes lactate and builds glucose back out of it, then returns the glucose to circulation, a loop that keeps carbohydrate in the system rather than losing it.

And lactate does a second job that has nothing to do with energy. It acts as a signal. Concentrations of it influence gene expression inside muscle, and the transporter that imports it is itself upregulated by endurance training. Some of the adaptation an endurance athlete accumulates over years is a response to the very molecule the folklore says they should be trying to avoid.

Which reframes what a high threshold actually is. An athlete with a high threshold is not somebody who produces less lactate. They are somebody who can clear far more of it, because they have more mitochondria to oxidise it, more capillaries to deliver and collect it, and more transporter protein to move it across membranes. The threshold is a clearance number wearing a production number's clothes.

1. 1Glycolysis outruns the mitochondria in a fast-twitch fibreGlucose is broken down to pyruvate faster than that fibre's mitochondria can take pyruvate in. The surplus pyruvate is reduced to lactate, a reaction that consumes a proton rather than releasing one.
2. 2Lactate leaves the fibre through a transporterExport is carried by a membrane transporter abundant in glycolytic fibres. Nothing about this step requires the fibre to be short of oxygen.
3. 3A neighbouring slow-twitch fibre imports itOxidative fibres carry a different transporter, on the cell membrane and on the mitochondria inside. They convert the lactate back to pyruvate and burn it.
4. 4What is not consumed locally enters the bloodstreamOnly the surplus reaches the circulation, which is the only place a finger prick can sample. Blood concentration is therefore the leftovers, not the output.
5. 5The heart, the brain and other muscles take it upCardiac muscle in particular uses lactate readily under load, and takes more of it as the supply rises.
6. 6The liver rebuilds glucose from the remainderLactate is a raw material for making glucose, which is returned to the blood. Carbohydrate that would otherwise be lost is recycled.
7. 7Blood concentration settles wherever supply and demand meetThe number on the analyser is the balance of two rates, not a measure of how much was made.

The route is the same at rest and at race pace. What changes with intensity is how much is moving and how quickly the consuming tissues can accept it.

What the number on the analyser actually is

A lactate meter reads a drop of capillary blood, usually from a fingertip or an earlobe. Understanding what that reading is, and is not, saves an enormous amount of misinterpretation.

It is a concentration in one compartment. Muscle lactate concentration can be several times blood concentration during hard work, and the gradient between them is what drives the transfer. Blood is the overflow channel, not the reservoir.

It is a balance of two rates. Concentration holds steady whenever appearance and disappearance are equal, and that can happen at a low value or a high one. Two athletes reading the same 3 mmol/L may be producing wildly different amounts and clearing wildly different amounts. The reading tells you the difference between the rates, not either rate.

It lags. When intensity steps up, blood lactate does not arrive at its new value instantly. It has to be produced, exported, carried by the circulation and mixed. This is why the stages of a step test have to be long enough for the value to approach a plateau, and why a two-minute stage produces a flatter, more flattering curve than a five-minute one at the same speeds.

It depends on where you take it from. Fingertip, earlobe and venous samples do not read alike, and the difference is not a constant offset that can be subtracted away. Analysers differ too, and so do the strips. None of this makes the measurement useless. It makes it a measurement that is only interpretable against itself: same athlete, same site, same analyser, same protocol.

That constraint is the single most common failure in amateur lactate testing. A test in March on an earlobe with one meter and a test in September on a fingertip with another does not measure progress. It measures two things that happen to share a unit.

Lactate threshold explained: there are two of them, and neither is a switch

The phrase "the lactate threshold", singular and definite, is the source of half the muddle. There are two commonly identified boundaries on the curve, and they mean different things.

The first, usually written LT1 and often called the aerobic threshold, is the intensity at which blood lactate first rises meaningfully above its resting value. Below it, an athlete is in genuinely comfortable territory: lactate barely moves, fat contributes a large share of the fuel, and the limit on duration is fuel and mechanical wear rather than metabolic stress. It is the top of easy running, and it is where a large fraction of a serious endurance athlete's training volume sits.

The second, LT2, goes by more names than any other quantity in the sport: anaerobic threshold, lactate turn point, onset of blood lactate accumulation, maximal lactate steady state, or simply threshold. It is the highest intensity at which blood lactate still stabilises. Go a little above it and the concentration does not settle at a new higher plateau, it climbs for as long as the effort continues.

"Anaerobic threshold" is the worst of those names and the most popular. Nothing becomes anaerobic. Oxygen delivery to working muscle is not the limiting factor at that intensity for most athletes, oxidative metabolism is supplying the overwhelming majority of the ATP, and no switch is thrown from one system to another. Energy pathways run continuously and in parallel, with their relative contributions sliding as intensity rises. The word survives because it is short and because it was in the textbooks first.

There is a deeper problem with the word threshold itself. A threshold suggests a discontinuity: a line with different physics either side. What the data actually give you is a smooth curve that gets steeper. Somebody has to decide where the interesting part of that curve is, and that decision is a modelling choice made by a human being, not a feature discovered in the athlete. Which is exactly why the detection methods disagree, as they very much do.

The three intensity domains, which is the more useful version

Drop the word threshold for a moment and think about behaviour instead. The two boundaries divide effort into three domains, each with its own signature, and the domains are the thing worth knowing.

Moderate, below LT1. Blood lactate sits near baseline and stays there. Oxygen uptake reaches a steady value within a couple of minutes and stops rising. Duration is limited by glycogen, by fluid, by heat and by the mechanical toll of the movement, not by any metabolic instability. This is the domain of long runs and easy miles, and the domain in which most of the clearance machinery is built.

Heavy, between LT1 and LT2. Lactate is elevated and stable. Oxygen uptake reaches a steady value, but only after a delayed extra rise on top of the fast response, so the true oxygen cost is higher than the pace would suggest. Efforts here are sustainable for a long time, but not indefinitely, and the sensation of holding them is one of controlled discomfort that does not get worse minute by minute.

Severe, above LT2. No steady state exists. Lactate climbs continuously. Oxygen uptake, if the effort lasts long enough, is dragged all the way up to maximum whether the pace called for it or not. Time to exhaustion in this domain is strikingly predictable: there is a fixed quantity of work available above the boundary, and the further above it you go, the faster you spend it.

The clean fact underneath is that this changes what a threshold number is for. It is not a target to hit in a session. It is the border between a workout that stabilises and a workout that has a clock running on it, and the practical difference between the two is enormous. Endurance training in athletics and everywhere else is largely an argument about how to distribute time across those three domains.

How a lactate threshold test is actually run

The standard protocol is unglamorous and it is very easy to run badly.

The athlete arrives rested and, importantly, normally fuelled. Glycogen state changes the curve: a depleted athlete produces less lactate at any given speed, so testing after a low-carbohydrate day returns a flattering result that reflects an empty fuel tank rather than an improved engine. Prior training in the previous day or two shifts it as well. Standardise those things or the test measures your week rather than your fitness.

Then a series of stages of increasing intensity, each three to five minutes long, with a fixed increment between them. Three minutes is common on a bike, four to five on a treadmill or a track, and the choice matters because it changes how close each value gets to its plateau. A blood sample is drawn at the end of each stage, either in the final seconds or in a break of about half a minute, along with heart rate and a rating of perceived effort. The stages continue until lactate is rising steeply, typically several stages past the point of interest so that there is enough curve on both sides to fit anything to.

What comes out is a set of paired values: speed or power against blood lactate, with heart rate alongside. Everything after that is arithmetic on those pairs.

1. 1Standardise the day beforeNormal carbohydrate intake, no hard session in the preceding 48 hours, no caffeine before the test unless caffeine is standard for every test. Fuel state moves the curve more than most people expect.
2. 2Rest sample and warm-upA baseline value is taken before work begins, then an easy warm-up long enough to settle heart rate and mechanics without generating lactate that contaminates the first stages.
3. 3Stage one, well below the expected thresholdStarting low costs a few minutes and buys the flat part of the curve. Several detection methods depend on having genuine baseline points to anchor to.
4. 4Hold each stage for three to five minutesLong enough for blood lactate to approach a plateau at that intensity. Shorter stages read low and produce a threshold that is too fast.
5. 5Sample at the end of every stageCapillary blood from the same site every time, plus heart rate and perceived effort. A break of roughly thirty seconds is acceptable and must then be used at every stage.
6. 6Increase by a fixed increment and repeatThe increment is chosen so that six to ten usable stages are collected. Too large an increment leaves too few points to fit a curve through.
7. 7Continue past the steep riseTesting stops when lactate is climbing sharply or the athlete cannot hold the next stage. Points above the turn are needed for the fitting methods to work.
8. 8Fit the curve and apply one detection methodThe chosen method returns a speed or power, and the heart rate recorded at that intensity. Which method you choose changes the answer, so it is fixed once and never varied.

The sequence is standard practice in laboratory and field testing. Stage length and increment vary by sport and by laboratory, and both change the answer, which is why they have to be held constant between tests.

Five ways to read one curve, five different answers

This is the part the popular explanations leave out, and it is the part that determines whether a threshold number is worth anything.

Take a constructed step test. The numbers below are invented, with round increments chosen so the reading is easy to follow. A runner completes seven stages, and the blood lactate values come out like this.

Stage speed, km/h	Blood lactate, mmol/L
11.0	1.0
12.0	1.1
13.0	1.4
14.0	2.0
15.0	3.2
16.0	5.4
17.0	8.6

The invented seven-stage test above, plotted. Nothing discontinuous happens anywhere on this curve, which is the whole difficulty: every detection method is a rule for choosing a point on a smooth line.

Show the numbers

Constructed example: the shape a step test produces

Item	Blood lactate
11.0	1mmol/L
12.0	1.1mmol/L
13.0	1.4mmol/L
14.0	2mmol/L
15.0	3.2mmol/L
16.0	5.4mmol/L
17.0	8.6mmol/L

Look at that line and the problem is obvious. There is no corner. The curve flattens at the bottom and steepens at the top and passes smoothly through everything in between, and any threshold you take from it is a point somebody decided to name.

Now the methods.

Fixed value. Read off the speed at which the curve crosses a prescribed concentration, most often 4 mmol/L. It is simple, it is comparable between athletes, and it ignores the individual entirely. Real athletes' steady-state concentrations vary a long way either side of four, so for some people the fixed value lands above their true sustainable intensity and for others below it.

Baseline plus a fixed rise. Take the resting or lowest value and add one millimole per litre. This respects individual baselines, which is an improvement, and it is highly sensitive to the accuracy of that baseline, which is not.

Log-log transformation. Plot the logarithm of lactate against the logarithm of intensity, where the early part of the relationship becomes close to a straight line, and find where it departs from that line. This tends to identify the first threshold rather than the second, and it is the most objective of the low-intensity methods.

Dmax. Draw a straight line between the first and last data points, fit a curve through all of them, and find the point on the curve furthest from that straight line. It is elegant, and it has a serious weakness: the answer depends on where the test began and ended. Add an easier first stage and the chord moves, and so does the threshold.

Modified Dmax. Same construction, but the chord starts at the first point showing a genuine rise rather than the very first stage. It removes some of the protocol dependence, and it typically returns a higher intensity than plain Dmax.

Log-log13.5km/h

Baseline plus 1 mmol/L13.9km/h

Dmax14.9km/h

Fixed 4 mmol/L15.4km/h

Modified Dmax15.7km/h

Read from the invented seven-stage test in the table above. The values are what each method returns from that curve, not any athlete's data. The point is the spread, which is large enough to change what a training session feels like.

Show the numbers

Constructed example: five detection methods applied to one step test

Item	Value
Log-log	13.5km/h
Baseline plus 1 mmol/L	13.9km/h
Dmax	14.9km/h
Fixed 4 mmol/L	15.4km/h
Modified Dmax	15.7km/h

Better than two kilometres an hour separates the extremes of that constructed set, and the pattern of which methods sit high and which sit low is a real property of the methods rather than an artefact of these particular numbers. Log-log and small fixed rises pick up the first threshold. Fixed 4 mmol/L and modified Dmax pick up something nearer the second. Plain Dmax lands somewhere in between and moves around with the protocol.

Two consequences follow. Comparing your threshold to somebody else's is close to meaningless unless you know their method, their stage length and their analyser. And tracking your own progress requires freezing every one of those choices, because a change of method will manufacture an improvement or a decline out of nothing.

Maximal lactate steady state, and the only definition with a behaviour behind it

There is one definition that is not a reading of a curve at all, and it is the one the others are trying to approximate.

Maximal lactate steady state is determined by doing the thing directly. The athlete performs a series of constant-intensity efforts, conventionally around half an hour each, on separate days. In each, blood lactate is sampled at intervals. If the concentration between the tenth minute and the thirtieth rises by no more than about one millimole per litre, that intensity qualifies as a steady state. The highest intensity that still qualifies is the maximal lactate steady state.

Notice what that definition does. It does not ask what concentration you reached. It asks whether the concentration was stable. An athlete whose steady state sits at 6 mmol/L and one whose steady state sits at 2.5 mmol/L are both, at their own boundary, doing the same physiological thing, and a fixed 4 mmol/L rule would misjudge both of them in opposite directions.

The catch is cost. Finding it properly takes four or five separate visits, each with a warm-up and a half-hour effort and a bracketing search. Very few athletes will do that. So the incremental step test survives as the practical stand-in, and every method described in the previous section is ultimately a cheap estimate of this expensive number.

Critical power and critical speed, arrived at from the other end

There is a second way to find the same boundary, and it needs no blood at all.

Set an athlete several maximal efforts of different durations, from a few minutes to twenty or so, with proper recovery between them. Plot power, or speed, against how long it was held. The relationship is a hyperbola, and it has two parameters. One is an asymptote, the intensity the curve flattens towards, called critical power on a bike and critical speed on foot. The other is a fixed quantity of work available above that asymptote, spent once and then gone until recovery replenishes it.

That model does something the lactate curve cannot: it predicts. Give it a target intensity above the asymptote and it returns how long that intensity can be held, from the size of the finite reserve divided by the rate you are overspending. Below the asymptote, the model says the effort is sustainable and the limit comes from elsewhere.

Critical power and the lactate markers agree well, which is the reassuring part. They are estimating the same boundary from opposite directions, one from metabolic behaviour and one from the shape of the fatigue curve. Published comparisons in the same riders find strong correlations across the various lactate markers, with the critical power estimate tending to sit towards the upper end of the range they produce rather than in the middle.

If you have a power meter and no lactate analyser, this is the better route, and it is closely related to what most riders already call their threshold. The distinction between the two, and the reason functional threshold power is a convenience rather than a physiological quantity, is worked through in the piece on what a power meter actually measures and what FTP approximates.

What actually raises the threshold

Because the threshold is a clearance number, the adaptations that raise it are the ones that increase the body's capacity to take lactate back out of circulation and to avoid producing the surplus in the first place.

Mitochondrial density and enzyme activity. More mitochondria in a fibre means more pyruvate can be accepted directly, so less of it spills over into lactate. It also means more capacity to oxidise lactate arriving from elsewhere. This is the largest single lever, and it responds to accumulated aerobic work.

Capillary density. New capillaries around a fibre shorten the diffusion distance for oxygen going in and for lactate coming out. They also increase the time blood spends in contact with the muscle, which helps both exchanges.

Transporter protein. The proteins that carry lactate across membranes are expressed in greater quantity in trained muscle. More transporter means the same concentration gradient moves more lactate, which is a straightforward increase in clearance capacity.

Recruitment and fibre character. Training shifts the pattern of which fibres do the work at a given pace and shifts the metabolic character of the fibres themselves towards the oxidative end. A pace that once required substantial fast-twitch recruitment eventually does not.

Delivery. Plasma volume, cardiac output and haemoglobin mass all determine how much oxygen arrives and how quickly lactate is carried away to the tissues that will consume it.

The training that produces these is less exotic than the marketing suggests. Large volumes of moderate work build mitochondrial and capillary density with very low fatigue cost. Work at and slightly above the second threshold provides a stimulus the easy work does not, and comes with a recovery bill that limits how much of it is useful. The long argument in endurance coaching about polarised distributions versus more time spent at threshold is an argument about the ratio, not about whether either ingredient is needed.

One timescale is worth stating plainly, because it changes how a serious athlete thinks about a decade of training. Maximal oxygen uptake tends to plateau relatively early in a training career. The threshold, expressed as a fraction of that maximum, keeps improving for years afterwards. That is the physiological reason distance runners peak later than the raw engine numbers would suggest.

Where the threshold sits in the model of endurance performance

The classical model of endurance performance has three inputs. Maximal oxygen uptake sets the ceiling on aerobic power. The fraction of that ceiling you can sustain, which is essentially the threshold, sets how much of it is usable for a long effort. Economy sets how much oxygen a given speed costs. Multiply the first two and divide by the third and you get a sustainable speed, which predicts a large share of the variance in distance running performance.

That framing is sound, but the intuitive weighting people put on it is not. Large-sample analyses that decompose the contributions find the maximal oxygen uptake term accounting for the greatest share of the differences in threshold speed between athletes, economy accounting for a substantial second share, and fractional utilisation, the threshold expressed as a percentage, accounting for the smallest of the three. The threshold speed is what predicts the race. The percentage on its own does much less work than its prominence in training talk implies.

This is worth sitting with, because it changes what "raise your threshold" means. Threshold speed can rise because the ceiling rose, because the percentage rose, or because the same speed became cheaper. Three different training problems wearing one label. The engine side is covered in the article on what maximal oxygen uptake does and does not tell a runner, and the cost side has become an equipment question as well as a physiological one, which is the subject of the piece on how carbon-plated shoes changed distance running.

The threshold is not a constant, even inside one race

A laboratory test gives you a number obtained on a fresh athlete. Races are not run by fresh athletes for very long.

After a couple of hours of sustained work, the whole lactate curve shifts. The intensity that produced a steady state at the start produces a slow climb by hour three, and the power or speed at which the threshold sits falls. The property being described has acquired a name of its own, durability, and it is increasingly treated as a fourth determinant of endurance performance alongside the classical three, precisely because two athletes with identical fresh numbers can differ enormously in what remains of them late on.

The practical implication for a marathon is direct: a threshold pace measured in a rested state overstates what will be available at thirty kilometres, and any pacing plan built on the fresh number without a discount is a plan to blow up on schedule. How to set the target properly is worked through in the article on pacing a marathon as a fuel problem rather than a fitness problem.

Several other things move the curve, and all of them are worth knowing before interpreting a test.

Glycogen availability. Low muscle glycogen suppresses lactate at every intensity, because there is less substrate to make it from. The curve shifts right, which looks like an improvement and is nothing of the kind.

Heat. Working in the heat raises lactate at a given intensity, partly through altered blood flow distribution and partly through changed fibre recruitment. A summer test and a winter test are not comparable.

Altitude. Reduced oxygen availability raises lactate at any given power in the short term, and prolonged exposure changes the response again.

Prior exercise, caffeine and even the previous meal. Each shifts the curve by a small amount, and small amounts matter when the methods themselves disagree by a couple of per cent.

Lactate threshold explained in training zones, and where the zones go wrong

Zone models are downstream of everything above, and they inherit every weakness in it.

The three-zone model maps directly onto the three intensity domains and is the honest version: below LT1, between the thresholds, above LT2. It has the advantage that every boundary means something behavioural. The five and seven zone models used by most training platforms subdivide those bands further, usually for convenience in prescribing sessions, and the extra boundaries are conventions rather than physiological events.

The structural problem is that most zone schemes are anchored to a single number, either a threshold estimate or a maximum heart rate, and then derived by percentage. Every error in the anchor propagates into every zone. If the threshold was found with a method that reads high, an athlete's "easy" pace is not easy, their long runs are quietly heavy, and the accumulated fatigue arrives without an obvious cause.

Heart rate anchoring adds problems of its own. Heart rate lags a change in intensity by a minute or more, so it is useless for controlling short intervals. It drifts upward across a long effort at constant work rate, so the zone that was correct in the first hour is not correct in the third. It responds to heat, dehydration, sleep and caffeine. Pace and power are external measures and do not have these faults, though they have their own: pace is corrupted by wind and gradient, and power is only available in some sports.

Which of these controls to trust varies by discipline. In race walking, where the rules constrain the technique itself, the sustainable intensity is bounded by what can be held without breaking a form that judges are watching, so an athlete's threshold pace and their legal pace are two different constraints that have to be trained together. In a time trial the control problem is different again, and the whole discipline is one of holding an intensity just under the boundary for as long as the distance requires, which is why pacing a time trial is its own skill.

What the finger prick cannot tell you

A short list, because knowing the limits is what stops the number being misused.

It does not measure muscle lactate, only what spilled into the blood. It does not measure production, only the net of production and clearance. It does not measure fatigue, which arises from a collection of causes including phosphate accumulation, calcium handling failure, glycogen depletion and central drive, in proportions that vary by intensity and duration. It does not measure how you will feel in the third hour. And it cannot be compared across analysers, sampling sites or protocols without introducing an error that is often larger than the effect being looked for.

None of that makes testing pointless. It makes the test a tool with a specific job: locating the boundaries between intensity domains for one athlete, repeatably, so that training can be distributed with intent rather than by feel alone.

Getting something usable out of it

If you are going to test, five things have to be nailed down before the first number means anything.

Fix the protocol. Same stage length, same increment, same starting intensity, same warm-up. Write it down and never change it, because a change of stage length alone can move the answer by more than a season of training.

Fix the hardware and the site. One analyser, one make of strip, one sampling site. Earlobe and fingertip are both fine, and mixing them is not.

Fix the detection method. Choose one, apply it to every test, and record which one you used. If you want a second opinion, apply a second method to every test rather than switching between them.

Standardise the state. Same time of day, same fuelling in the preceding day, same rest, same caffeine. A test after a low-carbohydrate week will hand you a better-looking curve than a test after a normal one, and it will be lying.

Keep the whole curve, not the derived number. The single threshold value discards nearly all of the information. The shape of the curve, the baseline, the speed at which the first rise occurs and the steepness of the climb after the turn all change with training, and they change at different rates. Two tests a season apart with identical threshold speeds can have very different curves, and the difference is the part worth reading.

Do those five things and a lactate test becomes what it should be: a repeatable map of where your own intensity domains sit. Skip any of them and you have bought an expensive number that tells you what week you had. The rest of the long-form explainers on how sport actually works run on the same principle, which is that the mechanism is more useful than the headline figure, and rather harder to get wrong.