The Disappearing Complete Game: Starters, Bullpens and Workload by Era
Teams completed 148 games in 1904 and the whole majors completed 29 in 2025. A decade-by-decade look at complete games, relievers used per game and starter workload, with the causes and what the numbers cannot tell you.
In 1904 the Boston Americans completed 148 of the 157 games they played. Their starting pitcher finished nearly every game he began, and the rest of the American League was not far behind: the Detroit Tigers completed 143, and in the National League the St. Louis Cardinals completed 146 and the Cincinnati Reds 142. In 2025 the thirty major-league clubs combined for 29 complete games in the entire season, an average of one per team, and thirteen of the thirty did not complete a single game. The Cincinnati Reds led the majors with four. The same statistic, a starting pitcher finishing what he started, went from the norm to a rarity in about a century.
This article follows that decline decade by decade using the historical record through the 2025 season, which counts complete games for every American League and National League club. It also shows what replaced the complete game: more pitchers per game, shorter starts and a bullpen that now covers close to forty percent of every contest. The causes are partly strategic, partly physical and partly statistical, and the numbers let us see which is which.
A complete game means one pitcher recorded every out for his team. It does not have to last nine innings: a road pitcher who loses 3-2 completes eight, because the home team does not bat in the ninth, and a rain-shortened game can be complete at five. A shutout is a complete game with no runs allowed. A no-hitter thrown by one pitcher is also a complete game; the story is in the history of no-hitters and perfect games.
When finishing was the job, 1876 to 1909
In the 1870s the average major-league club completed 47.9 games per season, and in the 1880s, as seasons lengthened, the figure rose to 103.2, which was 94 percent of the games played. In the 1890s it was 117.0, and in the American and National Leagues from 1901 to 1909 it was 117.9, about 79 percent of games on a 150-game-plus schedule. The pitcher was expected to finish what he began, and the exceptions were injury or a blowout. Pitching changes were rare: the league average for 1901 to 1909 was 1.24 pitchers per team in each game.
The workloads were enormous. In 1904 Jack Chesbro of New York completed 48 games, the most in a season since 1900 in the historical record; George Mullin of Detroit completed 42 and Cy Young, then 37 years old, completed 40. Cy Young's career total of 749 complete games, which still leads the record by 103 over Pud Galvin's 646, is a number that can no longer be approached. Walter Johnson finished with 531.
It is worth remembering what a pitcher did in those years. Runs were scarce, strikeouts were rare and the ball stayed in play, so a starter could pitch to contact for nine innings without the repeated high-effort pitches that modern hitters force. The historical numbers agree: the teams of 1901 to 1909 averaged 14.5 shutouts per season, the highest decade figure in the American and National League record.
The first decline: 1910 to 1949
The drop began in the 1910s. Clubs averaged 85.5 complete games per team-season, which was 57 percent of their games, and in the 1920s 76.5, or 50 percent. In the 1930s the figure was 68.6 and in the 1940s 66.0. The decline was steady rather than sudden, and it occurred while the number of pitchers used per game crept upward, from 1.24 in 1901 to 1909 to 1.60 in the 1910s, 1.80 in the 1920s and 2.00 in the 1940s.
Two forces were at work. The first was the live ball. Babe Ruth hit 54 home runs in 1920 when the next man in the American League hit 19, and the home run became a weapon for the whole order. That changed the risk of every pitch late in a game. The 1920 season produced 0.26 home runs per team-game, 4.36 runs per team-game and 2.95 strikeouts per team-game. The second force was the first specialist. Saves were not an official statistic until 1969, so the historical save totals for earlier years are reconstructed by the modern definition, but the reconstruction shows clubs averaging 4.2 saves in 1901 to 1909, 9.0 in the 1910s, 11.4 in the 1920s and 13.6 in the 1930s. Relievers were taking over more of the late innings.
Even so, the complete game was still routine. In 1946, the first year after the war, the Detroit Tigers completed 94 of 155 games, Bob Feller led the majors with 36, Hal Newhouser had 29 and a young Dave Ferriss had 26. A pitcher who started in the 1940s was still expected to be there at the end of an ordinary game.
The great fade: 1950 to 1989
The next decline was bigger. In the 1950s clubs averaged 51.8 complete games, 33.5 percent of their games. In the 1960s the figure was 40.7, and 25 percent. The 1970s held near that level, at 40.8. In the 1980s it fell to 24.4, or 15.6 percent. In one lifetime the complete game went from half the games to one in six.
There was a brief reversal in 1968, the Year of the Pitcher. Juan Marichal completed 30 games, Bob Gibson 28 and Denny McLain 28, and the San Francisco Giants as a team completed 77 of 163. The average club completed 44.9 that year, up from 36.8 in 1966. The sport responded for 1969 by lowering the mound and shrinking the strike zone, and the average club figure fell back to 40.9.
The most extreme result of the 1980s is the 1980 Oakland Athletics. Under manager Billy Martin, the club completed 94 games, nearly double the 48 of the second-place Milwaukee Brewers and more than twice the 42 of the Baltimore Orioles, who won 100 games. Rick Langford had 28 complete games, Mike Norris 24 and Matt Keough 20; the A's finished 83-79. For comparison, the Chicago Cubs completed only 13 games that year. Oakland's total was not a sign that the old way was coming back. The A's completed 60 games in the strike-shortened 1981 season, and since then no club has completed more than 48 in any year.
By the late 1980s the question was no longer whether a starter would finish, but how rarely. In 1989 the average club completed 18.6 games.
The collapse: 1990 to the present
The 1990s averaged 11.5 per team-season, the 2000s 5.8, the 2010s 3.5 and the 2020s through 2025 1.1. In 2000 the league leader was Arizona with 16; in 2019 the Chicago White Sox and Cleveland led with six each, and Shane Bieber and Lucas Giolito led the individual race with three apiece. In 2025 no pitcher completed more than two. Teams began to finish a season with no complete games at all in 2007, when Florida, Texas and Washington did so, and 96 team-seasons through 2025 have ended with a zero.
The table summarizes the trend for the American and National Leagues, one row per decade, with 2025 added at the end.
| Decade | CG per team-season | CG share of team games | Pitchers used per team game |
|---|---|---|---|
| 1901 to 1909 | 117.9 | 78.8% | 1.24 |
| 1910s | 85.5 | 56.6% | 1.60 |
| 1920s | 76.5 | 49.6% | 1.80 |
| 1930s | 68.6 | 44.6% | 1.90 |
| 1940s | 66.0 | 42.6% | 2.00 |
| 1950s | 51.8 | 33.5% | 2.30 |
| 1960s | 40.7 | 25.2% | 2.55 |
| 1970s | 40.8 | 25.3% | 2.46 |
| 1980s | 24.4 | 15.6% | 2.72 |
| 1990s | 11.5 | 7.4% | 3.34 |
| 2000s | 5.8 | 3.6% | 3.76 |
| 2010s | 3.5 | 2.1% | 4.09 |
| 2025 only | 1.0 | 0.6% | 4.29 |
The pitchers-used column is the sum of all pitcher appearances divided by team games. It rose from 1.24 to 4.29, so a modern game uses between three and four times as many pitchers as a game in 1905. The Federal League of 1914 and 1915 is left out of these rows, and so are the Negro Leagues, which the historical record through 2025 includes under other league codes for 1920 to 1948, and whose story is told in the history of the Negro Leagues.
Why the complete game vanished
The causes overlap, and it helps to take them one at a time.
The batter became more dangerous per pitch. Using the season sheets, a team in 1920 hit 0.26 home runs per game and struck out 2.95 times; in 1980 the figures were 0.73 and 4.80, and in 2019 they were 1.39 and 8.81. Runs per team-game barely moved: 4.36, 4.29 and 4.83. The game kept scoring about the same, but it was increasingly decided by strikeouts and home runs, both of which require a high-effort pitch. A pitcher facing a lineup full of players who can drive any mistake out does not throw 120 relaxed pitches. He throws 100 hard ones. For more on that trend, see the evolution of the strikeout.
The specialist arrived and multiplied. Saves per team grew from 4.2 in 1901 to 1909 to 13.6 in the 1930s, 21.7 in the 1950s, 31.0 in the 1960s, 36.5 in the 1980s and about 40 since the 1990s, which suggests that by the late 1990s nearly every club had a pitcher assigned to the last inning. The role came with the statistic: once the save became official in 1969 it created a clear job for the best reliever.
Pitch counts and injury caution. Clubs began tracking pitches and treating a starter's workload as a resource to be conserved. The historical data cannot show pitch counts, but it shows the result. Among pitchers who made at least 10 starts and started at least 90 percent of their appearances, the average outing was about 8.5 innings per start in 1901 to 1909, 7.7 in the 1920s, 7.2 in the 1950s, 6.6 in the 1980s, 6.1 in the 2000s and 5.5 in 2025. That is an approximation, since it uses innings per start for full-time starters, but the direction is unmistakable.
Analytics. Teams learned that hitters do better each time they see a pitcher in a game, and that a fresh reliever often has the advantage over a tiring starter. That research gave managers a reason to remove a pitcher who was still effective, which is why modern starters are often pulled in the sixth or seventh inning while the score is close.
What the arithmetic looks like today
Consider a typical modern game. A starter who pitches 5.5 innings leaves 3.5 innings for the bullpen, which at about 1.2 innings per reliever means three relievers to cover the rest. Add the starter and the average is four pitchers per team per game, which matches the 4.29 in the table. In 1905 a starter averaged about 8.5 innings, leaving half an inning, which is why the figure was 1.24: one pitcher in most games and a second in about a quarter of them.
The same arithmetic explains why clubs carry so many pitchers. A club that uses more than four pitchers per game over 162 games needs a deep supply of arms who can pitch on consecutive days, and a steady flow of replacements from the minors when they are hurt. That is a cost that complete games would have avoided.
Shutouts changed hands
One quiet detail in the historical team totals shows how completely the bullpen has taken over. Team shutouts, which count games in which a club allowed no runs, barely moved while complete games disappeared. Clubs averaged 10.2 shutouts per season in the 1950s, 12.5 in the 1960s, 8.1 in the 1990s, 10.3 in the 2010s and 11.0 in 2025. In 2025 the average club threw 11.0 shutouts and 1.0 complete games, which means that nearly every shutout was a combined effort by a starter and several relievers.
That matters for how a reader should interpret old and new numbers. A shutout in 1905 was a single pitcher's feat, and a team record for them was a measure of one man's skill. A shutout today is a team statistic, and the team that records it may have used five pitchers. The score on the board is the same; the person responsible for it has changed. Counting a pitcher's own shutouts has become a much rarer exercise, while the team total has stayed in the same range.
Two aces, two styles
Two of the pitchers in the historical record show the shift in a single career. Nolan Ryan finished with 324 wins, 773 starts and 222 complete games, which is about 29 percent of his starts, along with 5,714 strikeouts. Greg Maddux, who pitched from 1986 to 2008, won 355 games in 740 starts with 109 complete games, about 15 percent, and 35 shutouts. Maddux was famous for efficiency, getting outs on few pitches, and he completed half as large a share as Ryan did. The fall in the complete game was not about the quality of the pitcher; it was about the era in which he worked.
Common questions
Who has the most career complete games? Cy Young, with 749, followed by Pud Galvin with 646 and Tim Keefe with 554. Walter Johnson is fifth with 531.
Has any modern pitcher come close to those totals? No. Of pitchers who debuted in 1990 or later, Roy Halladay leads with 67. Among pitchers who appeared in 2025, Justin Verlander has the most with 26, followed by Clayton Kershaw with 25.
Is a complete game still valuable? It still signals an unusually efficient outing, and a few fantasy formats score it, but most managers now weigh it less than avoiding the third time through the order.
Will the complete game come back? Nothing in the data suggests so. Almost every decade since the 1920s has had fewer complete games than the one before it, and the 1970s, level with the 1960s, were the only pause. If clubs found that a rule change, such as limits on roster size or pitcher usage, favored longer outings, the numbers would change, but that is speculation, not a trend the record supports.
Where to go next
The best companion to this article is The Evolution of the Strikeout, which tracks the other side of the same change. For the rarest kind of complete game, see No-Hitters and Perfect Games, and for current pitching leaders use the leaders page. The stats and analytics topic page collects current coverage.