Quality Start in Baseball: The 6-Inning, 3-Run Standard
A quality start in baseball is six or more innings with three earned runs or fewer. What MLB pitchers must do to earn one, how it is lost, and what it hides.
By CricketTaken EditorialPublished Baseball Stats & Positions26 min read
- Minimum innings
- 6
- Maximum earned runs
- 3
- Coined
- 1985, by John Lowe
- Single-season streak record
- 25, Framber Valdez in 2022
The starter walks off after the sixth with ten hits on his line, three runs across and his team behind by one. The bullpen takes over. Whatever happens from there, every stat service in the United States will credit him with a quality start.
A quality start in baseball is a start in which the pitcher completes at least six innings and allows three earned runs or fewer. Both halves must be true when he leaves the game. Unearned runs do not count against him, and the result is irrelevant: he can win, lose or get no decision. Major League Baseball describes the statistic in its glossary and it appears on every pitcher's stat page, but it is not in the Official Baseball Rules and no official scorer awards one.
What a quality start in baseball is: six innings, three earned runs
Two numbers, both thresholds. The innings figure is a minimum and the runs figure is a maximum.
Six innings means 18 outs recorded with the starter on the mound. Five and two-thirds is not enough, even if he has allowed nothing. A pitcher removed with two out in the sixth and a shutout going has not made a quality start; a pitcher who gets the eighteenth out with three runs in has.
Three earned runs is the ceiling. Zero, one, two and three all qualify. Four does not, whether he pitched six innings or nine.
The word "start" carries a condition of its own. Only the pitcher who throws his team's first pitch can earn one. A reliever who enters in the second inning and works six strong innings behind an opener has done the same job and gets nothing in this column.
| Pitching line | Quality start? | Why |
|---|---|---|
| 6 innings, 3 earned runs | Yes | Exactly at both thresholds |
| 5⅔ innings, 0 runs | No | One out short of six innings |
| 7 innings, 4 earned runs | No | One earned run over the limit |
| 6 innings, 3 earned runs, 4 unearned | Yes | Unearned runs are ignored |
| 9 innings, 3 earned runs | Yes | Length beyond six changes nothing |
| 8 innings, 4 earned runs | No | Fails on runs however long he lasted |
| 5-inning complete game, 0 runs, rain-shortened | No | The game was official; the six innings were never pitched |
| 6 innings in relief, 1 earned run | No | He did not start |
The seventh row comes up every summer. A game called for weather after five innings is a regulation game, and the starter gets a complete game and perhaps a shutout. He cannot get a quality start. The same logic applied to the seven-inning doubleheader games MLB played in 2020 and 2021: the bar stayed at six innings, which in those games meant all but one. How a shortened game becomes official is covered in our guide to how many innings a baseball game has.
As for "kept everywhere and official nowhere," the distinction is real. The Official Baseball Rules devote Rule 9 to scoring. It defines the win, the loss, the save and the shutout, and it instructs the official scorer how to credit each. The quality start is absent. Nobody in the press box rules on it. It is a product of the box score, computed afterward by whoever is keeping the database, and MLB's own site, the major stat references and every broadcaster all compute it the same way. In stat tables it is abbreviated QS.
The fine print: unearned runs, the decision and what goes unmeasured
Start with the adjective in "three earned runs." The scoring rules separate runs a pitcher is held responsible for from runs that scored because a fielder made an error or the catcher allowed a passed ball. Only the first kind count here.
This produces lines that look absurd. A starter can give up seven runs in six innings and be credited with a quality start, provided the official scorer judges that four of them would not have scored without a misplay. It does happen. An error with two out, followed by a string of hits, makes every run that follows unearned. The reasoning behind that reconstruction is laid out on our page comparing the earned run with the unearned run.
Because the statistic leans on the scorer's judgment, it can change after the game. A scorer may revise a call within 24 hours, and clubs can ask the league to review one. An error changed to a hit can convert an unearned run into an earned one, and with it a quality start into an ordinary outing. The opposite change can award a quality start days after the fact.
Next, what the definition ignores.
- The decision. A win, a loss and a no-decision are all compatible with a quality start. The statistic was built specifically to be indifferent to them.
- The score. His team can be shut out. It can win 14-2. The pitcher's line is all that is examined.
- Hits and walks. Twelve hits and five walks in six innings is a quality start if only three earned runs came of it.
- Pitch count. Six innings on 115 pitches counts the same as six on 68.
- Strikeouts. None are required.
- How he left. A pitcher pulled after six with the bases empty and one pulled mid-inning in the seventh are treated alike, so long as the final line meets the test.
The last point deserves emphasis, because it is where most confusion lies. There is no checkpoint at the end of the sixth. The test is applied once, to the final pitching line, after the last runner he was responsible for has either scored or been stranded.
One thing the definition does not ignore is the starter's own fielding. A pitcher who throws a bunt into right field has committed an error like any other fielder, and runs that follow from it are unearned to him. He can, in effect, protect his own quality start with a bad throw. Nobody designs a statistic to do that. It is a side effect of borrowing the earned run.
Losing one late: the seventh inning and the runners left behind
Six innings, three earned runs, 88 pitches. The manager sends him back out.
From that moment the starter has nothing to gain in this particular column and everything to lose. He already has a quality start in hand. A scoreless seventh leaves it in place. One more earned run erases it.
- Through sixThe starter has recorded 18 outs and been charged with no more than three earned runs. At this instant his line qualifies.
- Back outHe begins the seventh. Any earned run he allows from here is added to the same line.
- Pulled mid-inningIf he leaves with runners on base, those runners remain his responsibility until they score or the inning ends.
- BullpenA reliever who lets an inherited runner score adds that run to the starter's total, not his own.
- Final lineThe test is applied once, at the end. Six or more innings and three or fewer earned runs is a quality start. Four earned runs is not.
The sequence for a starter who has completed six innings with three earned runs or fewer. Runs scored by runners he leaves on base are charged to him under Rule 9.16 of the Official Baseball Rules.
The cruelest version involves the bullpen. A starter with two earned runs through six gives up a single and a walk to open the seventh and is lifted. He has allowed two runs. He is sitting in the dugout with a towel around his neck when the reliever serves up a double, both runners score, and his line becomes six-plus innings and four earned runs. The quality start is gone and he did not throw the pitch that lost it.
Runs are charged that way because the scoring rules assign each run to the pitcher who put the runner on base. The mechanics and their effect on a pitcher's earned run average are explained in our guide to what ERA measures.
How much cushion a starter carries into the seventh depends on his line through six.
| Earned runs through six innings | Earned runs he can still allow | Position |
|---|---|---|
| 0 | 3 | Comfortable; a three-run seventh still qualifies |
| 1 | 2 | Room for a two-run homer |
| 2 | 1 | One mistake allowed |
| 3 | 0 | Any earned run ends it |
| 4 or more | None | Already lost; nothing later restores it |
The bottom row is the other half of the rule. A quality start cannot be won back. A starter who gives up four in the first inning and then retires the next 24 batters has pitched magnificently for eight innings and has no quality start.
There is a mild incentive problem buried here, and it is fair to say so. A starter at six innings and three runs helps his statistics by coming out and can only hurt them by continuing. No serious evidence suggests managers or pitchers act on that. Managers decide by pitch count, matchups and the state of the bullpen. But the statistic does reward stopping at exactly the minimum, which is an odd thing for a measure of quality to do.
The 4.50 objection and why it misses
Six innings, three earned runs. Multiply three by nine, divide by six, and the earned run average for the game is 4.50.
That figure is the basis of the oldest complaint about the statistic. A 4.50 ERA is mediocre in nearly any season and poor in a low-scoring one. Critics, the broadcaster Tim McCarver prominent among them, argued for years that a standard calling such an outing "quality" was no standard at all.
The arithmetic is right and the conclusion is wrong.
Six and three is the floor of the category, not its center. A quality start is any outing at or above that line, and most of them are well above it: seven innings and one run, eight innings and two, a complete-game shutout. Judging the whole class by its weakest possible member is like judging everyone who passed an exam by the lowest passing mark.
The numbers bear this out. A study covering every quality start in the major leagues from 1984 through 1991 found that pitchers' combined ERA in those games was 1.91. Not 4.50, not 3.50. Under two. Later counts have landed in the same neighborhood, generally close to 2.00.
The bare-minimum line itself, exactly six innings and exactly three earned runs, turns out to be uncommon. It makes up a small minority of all quality starts in every count that has been published. The far more typical quality start is longer, or stingier, or both.
There is a second answer to the objection, and it is about what the alternative was. Before the statistic existed the only start-by-start summary a pitcher had was his decision. Six innings and three runs may be an ordinary night, but it is one in which the starter has taken his team two-thirds of the way through the game with the score within reach. Whether the team then wins is up to the hitters and the relievers.
The criticism does land in one respect. A starter who pitched nothing but minimum quality starts would have a 4.50 ERA and a quality start percentage of 100. No such pitcher has existed. But the possibility shows that the count alone cannot rank pitchers; it sorts outings into acceptable and not, and leaves finer judgments to other numbers.
What a quality start shows that a win does not
The pitcher's win is a team result pinned on one man. To earn it under Rule 9.17 a starter must pitch at least five innings, leave with his team ahead, and have the lead survive to the end. Two of those three things are out of his hands.
Run support decides the first. A starter who allows one run in eight innings loses if his team is shut out. Another who allows six in five innings wins if his team scores nine.
The bullpen decides the second. A lead handed over in the seventh and blown in the ninth takes the starter's win with it, however well he threw. The reliever who gave it up may even be credited with the victory if his team rallies. The rules for crediting the pitcher who finishes are in our explainer on the save.
The quality start strips both away. It asks only what the starter did while he was on the mound.
Two terms have grown up around the gap between the two measures. A tough loss is a loss charged to a pitcher in a game where he made a quality start. A cheap win is a win credited in a game where he did not. A pitcher's totals of each say more about his teammates than about him.
| Pitcher and season | Won-lost record | ERA | What the record concealed |
|---|---|---|---|
| Bob Gibson, 1968 Cardinals | 22-9 | 1.12 | Nine losses with the lowest ERA of the live-ball era |
| Nolan Ryan, 1987 Astros | 8-16 | 2.76 | Led the National League in ERA and strikeouts with a losing record |
| Felix Hernandez, 2010 Mariners | 13-12 | 2.27 | Led the American League in ERA; won the Cy Young Award |
| Jacob deGrom, 2018 Mets | 10-9 | 1.70 | Led the majors in ERA; won the Cy Young Award |
DeGrom's 2018 is the plainest case on record. He made 32 starts and delivered a quality start in 28 of them, and his team won fewer than half the games he pitched. A voter looking only at ten wins would have passed him over. The voters did not, and that Cy Young ballot, along with Hernandez's eight years earlier, marked the point at which the win stopped being the first thing anyone consulted.
The quality start did some of that work. It gave writers a simple count to set against the won-lost record, one that needed no explanation beyond a sentence. More sophisticated tools have since overtaken it. The approach described in our explainer on what WAR measures goes much further in separating a pitcher from his surroundings. But the quality start was the first widely used statistic to make the argument that a starter's job and his team's result are different things.
How often six innings and three runs is enough
The practical test of a threshold is whether meeting it wins games. On that test the quality start holds up well.
Counts over many seasons have found that a team whose starter delivers a quality start wins about two games in three. The exact figure moves with the scoring environment and from year to year, generally sitting somewhere between .650 and .700 as a winning percentage. That is the pace of a club that wins 105 to 113 games over a full season.
The reason is not mysterious. A team in a normal year scores somewhere between four and five runs a game. A starter who has allowed three or fewer through six innings has, on an average night for his offense, left with a lead or a tie. The bullpen has nine outs to get.
The other side of the ledger follows by arithmetic. Every game has a winner and a loser, so the league as a whole wins exactly half its games. If teams win two-thirds of the time when they get a quality start, they must win well under half the time when they do not, and the fewer quality starts there are, the closer that second figure creeps toward .400. A staff that produces them consistently is playing a different season from one that does not.
Two cautions belong here.
First, the average includes all the excellent starts. The team winning percentage behind a seven-inning, one-run outing is far higher than behind a bare six-and-three. Nobody should read "two in three" as the odds attached to the minimum line. For that line alone the figure is a good deal lower, as a 4.50 ERA for the night would predict.
Second, the relationship has shifted as the game has. When Lowe set the threshold in 1985, a starter who went six innings handed the last third of the game to a bullpen of three or four relievers of mixed ability. Today a starter who goes six is handing it to specialists throwing in the upper nineties, one inning each. Six innings is worth relatively more now than it was. The threshold has not moved, but what it buys has.
That change cuts against the old complaint. In the 1980s six innings was a slightly short outing. Now it is a long one.
Quality start percentage: the rate behind the count
A raw total of quality starts rewards showing up. A pitcher who makes 33 starts has more chances than one who makes 22. The rate corrects for that.
Quality start percentage = quality starts ÷ games started
Sixteen quality starts in 30 outings is 53 percent. Twenty-four in 30 is 80 percent. The first describes a useful mid-rotation starter; the second, one of the best pitchers in the league.
- 96%Greg Maddux, 1994
- 94%Dwight Gooden, 1985
- 70%Tom Seaver, career
Greg Maddux made a quality start in 24 of 25 outings in the strike-shortened 1994 season; Dwight Gooden in 33 of 35 in 1985; Tom Seaver in 454 of 647 across his career. Rounded to the nearest whole percent.
Those are the outer limits. Maddux failed to make a quality start once in 1994. Gooden, at 20 years old, missed twice in 35 tries. Over a whole career the standard falls, because careers include the early struggles and the long decline, and Seaver's 70 percent across 647 starts is about as high as any pitcher with a full modern career has managed.
As a working scale for a single season: a starter above 70 percent is an ace, one between 55 and 70 is a front-of-the-rotation arm, and one below 40 is not giving his team length or is giving up too many runs when he does.
The league rate has fallen
The proportion of all major league starts that qualify has dropped sharply. In the early 2010s more than half of all starts were quality starts. By the early 2020s the share was closer to one in three, with small recoveries in some seasons.
Pitchers did not suddenly get worse. What changed most is the innings half of the definition. A starter removed after five and a third with one run allowed has pitched well and fails the test. There are far more of those outings than there used to be.
This makes comparisons across eras awkward. A 60 percent rate in a year when the league manages 35 is a more impressive thing than a 60 percent rate in a year when the league manages 53. Anyone setting a current pitcher's percentage beside one from the 1980s should adjust in his favor.
Why quality starts are harder to come by
In the middle of the 1970s major league starters finished what they began more than a thousand times a season. In recent years the whole sport has produced a few dozen complete games annually. The quality start has not collapsed like that, but the same forces are squeezing it.
The pitch count. Somewhere around 100 pitches, nearly every organization now takes the ball. The figure is a convention more than a medical finding, but it is observed almost universally, and it is reached sooner than it used to be. Hitters take more pitches and strike out more, and strikeouts cost more pitches than ground balls. A starter who needs 17 or 18 pitches an inning is at 100 before he completes the sixth.
The third time through the order. Hitters do better against a starter each time they face him in a game. The effect is well documented and large by the third trip, and front offices have responded by pulling starters as the top of the lineup comes up again, which tends to fall in the fifth or sixth inning. A starter cruising with one run allowed through five and a third is often removed on that basis alone.
The opener. In May 2018 the Tampa Bay Rays began using a reliever to pitch the first inning before bringing in a longer pitcher, so that the pitcher expected to carry the game would miss the top of the order once. The opener cannot record a quality start; he faces a handful of batters. The pitcher who follows him cannot either, because he did not start. Every game played this way is one in which the statistic cannot be earned.
The five-inning start. Even without an opener, many clubs now plan for five innings from the back of the rotation and treat anything more as a bonus. Five innings qualifies a starter for a win. It does not qualify him for a quality start.
The bullpen behind him. This is the cause underneath the others. Managers pull starters early because the alternative is so good. A fresh reliever throwing 98 miles an hour for one inning is, on average, a better bet than a tiring starter facing hitters for the third time. The way clubs stack those relievers is described in our guide to the bullpen.
The average start across the majors lasted about six innings for decades and was still just under that mark in the middle of the 2010s. By the end of that decade it was barely over five. When the typical outing ends before the sixth inning is complete, the typical outing is not a quality start no matter how few runs score.
Rule changes aimed at pace, among them the pitch timer covered in our piece on the pitch clock, have not reversed this. MLB has discussed measures intended to keep starters in games longer. Through the 2026 season none had been adopted at the major league level.
What the quality start count hides
It is a yes-or-no statistic, and yes-or-no statistics throw information away.
A complete-game shutout with 12 strikeouts and a wobbly six innings with three runs across go into the season total as one each. So do a gem and a grind. Two pitchers with 20 quality starts apiece may have had very different years, and the count gives no hint which was better.
The cutoffs create cliffs. One out or one run either side of the line flips the result, and the pairs below show how arbitrary that can look.
| Outing that counts | Outing that does not | The better performance |
|---|---|---|
| 6 innings, 3 earned runs (4.50 ERA) | 9 innings, 4 earned runs (4.00 ERA) | The second, by ERA and by length |
| 6 innings, 3 earned runs | 5⅔ innings, 0 runs | Arguably the second |
| 6 innings, 3 earned and 5 unearned runs | 6 innings, 4 earned runs | The second; four total runs against eight |
| 6 innings, 11 hits, 5 walks, 3 earned runs | 7 innings, 3 hits, 0 walks, 4 earned runs | Unclear, which is the point |
The first row is the well-known paradox. A complete game with four earned runs is a better night's work by nearly any standard than six innings with three. One is a quality start. The other is not.
Beyond the cliffs, several things go unrecorded.
Strikeouts earn nothing and walks cost nothing. A starter who strands eleven runners has the same quality start as one who faced the minimum. The traffic he allowed, which a statistic like WHIP is designed to count, does not enter into it.
Context is absent. Three runs in six innings at altitude in Denver is a better performance than the same line in a pitcher's park at sea level. Three runs in 1968, when the league scored about three and a half a game, was a below-average start. Three runs in 2000 was a good one. The definition applies one line to all of them.
The quality of the opposition is ignored, and so is the weather, the umpire and the defense.
And the unearned-run exemption brings in the scorer's judgment. Two identical innings can produce an earned run in one ballpark and an unearned run in another, depending on whether a hard-hit ground ball was ruled a hit or an error.
None of this is a secret, and none of it makes the statistic useless. It was designed as a blunt count of acceptable outings. It does that. Asking it to rank the acceptable ones is asking for something it was never built to give.
Stricter yardsticks: seven innings and Bill James's game score
The complaints produced alternatives. Most tighten one threshold or both.
The best known is the high quality start, a version often associated with Nolan Ryan: seven innings or more with three earned runs or fewer. Others have proposed seven innings and two runs, or eight innings and one run of any kind under names such as "dominant start." Each fixes the 4.50 problem. Seven innings and three runs is a 3.86 ERA; seven and two is 2.57.
None has caught on. The stricter versions are rarer by design, and as starters have thrown fewer innings they have become rarer still, to the point where a seven-inning standard would now exclude most of what even very good pitchers do. A tougher bar also does nothing about the underlying limitation. It is still a line, with a cliff on each side.
Game score
Bill James took a different route. In his 1988 Baseball Abstract he introduced game score, which replaces the yes-or-no verdict with a number. He described it, with typical modesty, as something to play around with. It has outlasted the remark.
Every start begins at 50 points. From there:
- Add 1 point for each out recorded.
- Add 2 points for each inning completed after the fourth.
- Add 1 point for each strikeout.
- Subtract 2 points for each hit allowed.
- Subtract 4 points for each earned run.
- Subtract 2 points for each unearned run.
- Subtract 1 point for each walk.
An average start comes out near 50. A score above 70 is a strong outing, and anything in the 90s is the best game a pitcher will throw in a season.
Apply it to a bare-minimum quality start: six innings, six hits, three earned runs, two walks, four strikeouts. Fifty, plus 18 for the outs, plus 4 for completing the fifth and sixth, plus 4 for the strikeouts, makes 76. Subtract 12 for the hits, 12 for the runs and 2 for the walks. The result is 50. James's formula rates the minimum quality start as an average outing, which is about right.
Game score answers nearly every objection raised in the previous section. Strikeouts count. Walks and hits cost. Unearned runs are penalized at half weight, not waved away. A nine-inning, four-run complete game outscores a six-inning, three-run start, as it should.
A revised version by the analyst Tom Tango, which starts at 40 and adds a penalty for home runs, is the one MLB's own glossary now sets out. Two formulas under one name is a small confusion worth knowing about when figures from different sources disagree.
What game score lacks is the thing the quality start has: a one-line definition anyone can apply from the stands.
Streaks: Gibson, deGrom and Framber Valdez's 25 in a row
Consistency is the one thing a yes-or-no statistic measures better than an average does, and the streak records are where the quality start is at its most persuasive.
Bob Gibson of the St. Louis Cardinals made 26 in a row from September 12, 1967, to July 30, 1968. Nobody called them quality starts, since the term was 17 years away. The streak was identified later, when researchers ran the definition back through old box scores. Most of it fell in the season in which he posted a 1.12 ERA.
Jacob deGrom of the New York Mets equaled the mark half a century later, with 26 from May 18, 2018, to April 3, 2019. The first 24 came in 2018 and closed out his season, which at the time was a record for consecutive quality starts within one year.
Framber Valdez of the Houston Astros broke that single-season record in 2022. From April 25 through September 18 he made 25 consecutive quality starts. He came up one short of the all-time mark.
Valdez's streak says something about the statistic. He was not, that year, the most overpowering pitcher in his league. He was a left-handed sinkerballer who got ground balls by the dozen, rarely had a short outing, and finished with a 2.82 ERA across more than 200 innings. Several of the 25 were of the six-inning, two-or-three-run kind. What he did not do, for nearly five months, was have a bad night. Every fifth day the Astros got at least six innings with the game in reach, and their bullpen got the rest it needed. Houston won the World Series that fall, with Valdez the winning pitcher in two of its four victories.
That is what a streak of this kind captures. It rewards the floor of a pitcher's performance, not the ceiling. A pitcher can throw a one-hit shutout and follow it with a four-inning disaster, and his streak stands at zero.
The streak records are also among the safest in the book for the reasons given above. With the league-wide rate near one in three, a run of 25 requires a pitcher to do something well above the average, without interruption, for most of a season. Durable starters capable of it are scarcer than they were. A related and rarer feat, finishing a game without allowing a hit, is covered in our comparison of the perfect game and the no-hitter.
Where the quality start came from: John Lowe, 1985
The term first appeared in The Philadelphia Inquirer in 1985. John Lowe, then covering baseball for the paper, wanted a way to say how often a starting pitcher had done his job, and he did not think the won-lost record said it.
The problem he was addressing is familiar to anyone who follows one team every night. Starters pitch well and lose, or pitch badly and win, and the number in the next morning's paper that summarizes their season is the one most distorted by those accidents. Lowe proposed counting something else: the games in which the starter gave his team a real chance.
Six innings and three earned runs was a judgment about the game as it was played in the mid-1980s. Starters then averaged a little over six innings an outing. League ERAs sat a shade under 4.00. A starter who met both marks had done slightly better than par on length and on runs, and had left his bullpen three innings to cover.
Lowe did not hold out the statistic as a measure of excellence, and much of the later criticism argued against a claim he never made. The word "quality" probably oversold it. "Adequate start" would have been more accurate and would never have caught on.
It spread quickly. Other newspapers adopted it within a few seasons, the commercial stat services that supplied box scores began to carry it, and by the 1990s it was a fixture of pitcher profiles and game notes. Researchers with access to complete play-by-play records extended it backward through the twentieth century, which is how streaks and totals for pitchers who retired long before 1985 came to exist.
Lowe went on to a long career covering the Detroit Tigers for the Detroit Free Press. The statistic outgrew its origin, as these things do. Forty years on it sits in an odd position: too crude for front offices, which have far better tools, and still the quickest honest answer to the question fans ask after a game. Did the starter do his job?
For the rest of the pitching line and the rules behind it, our baseball section collects the related explainers.