Bob Gibson had a 1.12 earned run average in 1968. Pedro Martinez had a 1.74 in 2000.
Gibson's is the lower number and Martinez had the better season, and the reason is one column that almost never gets printed next to either of them. The National League in 1968 averaged 2.99 earned runs per nine innings. The American League in 2000 averaged 4.92.
Gibson was 1.87 runs better than his league. Martinez was 3.18 better than his. That is the largest gap between a qualified starter and his league in the history of the sport, and it is not close.
Why the raw list is useless
Rank every qualified season by earned run average and this is what you get:
| Pitcher | Season | ERA | |---|---|---| | Dutch Leonard | 1914 | 0.96 | | Mordecai Brown | 1906 | 1.04 | | Bob Gibson | 1968 | 1.12 | | Christy Mathewson | 1909 | 1.14 | | Walter Johnson | 1913 | 1.14 | | Jack Pfiester | 1907 | 1.15 | | Addie Joss | 1908 | 1.16 | | Carl Lundgren | 1907 | 1.17 | | Grover Alexander | 1915 | 1.22 | | Cy Young | 1908 | 1.26 |
Nine of those ten seasons came between 1906 and 1915, in a period when the ball was soft, the same ball stayed in play until it came apart, and nobody hit home runs. Carl Lundgren's 1.17 in 1907 is a fine season. It is not one of the ten best in history, it is just a fine season played in the lowest-scoring environment the sport has ever had.
The list is measuring the era, not the pitcher.
The list that adjusts for it
Divide each man's earned run average into his own league's, and the picture changes completely:
| Pitcher | Season | ERA | League ERA | Index | |---|---|---|---|---| | Dutch Leonard | 1914 | 0.96 | 2.73 | 284 | | Pedro Martinez | 2000 | 1.74 | 4.92 | 283 | | Greg Maddux | 1994 | 1.56 | 4.22 | 271 | | Bob Gibson | 1968 | 1.12 | 2.99 | 266 | | Greg Maddux | 1995 | 1.63 | 4.18 | 257 | | Walter Johnson | 1913 | 1.14 | 2.93 | 256 | | Mordecai Brown | 1906 | 1.04 | 2.62 | 253 | | Walter Johnson | 1912 | 1.39 | 3.34 | 240 | | Jacob deGrom | 2018 | 1.70 | 4.03 | 237 | | Zack Greinke | 2015 | 1.66 | 3.91 | 236 | | Pedro Martinez | 1999 | 2.07 | 4.87 | 236 | | Dwight Gooden | 1985 | 1.53 | 3.60 | 235 |
An index of 200 means a pitcher allowed half as many earned runs per inning as his league. Every season here is above 235.
Now the deadball era holds four places instead of nine, and the modern seasons that belong are there: Martinez twice, Maddux twice, deGrom, Greinke, Gooden.
Pedro Martinez, 2000
A 1.74 earned run average, 284 strikeouts, 217 innings, and an 18 and 6 record for Boston.
The context is everything. 2000 was the highest-scoring season of the whole steroid era. The average American League team scored 5.29 runs a game. Pedro pitched in Fenway Park, in the division with the Yankees lineup that won three championships in four years, and gave up 1.74.
The next best qualified earned run average in the American League that year belonged to Roger Clemens, at 3.70. Pedro was not leading his league, he was two full runs clear of the second best pitcher in it, and the second best pitcher in it was Roger Clemens.
He won the Cy Young award. He also won it the year before, in 1999, when he struck out 313 men and walked 37, a ratio of 8.46 strikeouts per walk, while posting a 23 and 4 record.
Those two seasons together are the best consecutive pair any pitcher has had in the live-ball era.
Bob Gibson, 1968, and what happened next
Gibson's 1.12 across 304 innings is still the lowest of any pitcher with 300 innings in the live-ball era, and he won both the Cy Young award and the Most Valuable Player award for it.
He also went 22 and 9. In a season with a 1.12 earned run average, he lost nine times.
And his 1.12 is not merely the lowest since 1920, it is more than half a run clear of second place. Carl Hubbell's 1.66 in 1933 is the nearest anyone else has come with 300 innings.
1968 is remembered as the Year of the Pitcher because everybody was doing some version of this. The two leagues combined hit .237. Scoring fell so far that the following winter the rules were changed: the mound was lowered and the strike zone was cut back, deliberately, to give hitters the game back.
So Gibson's season is both one of the greatest ever pitched and the final argument in a case for changing the rules. Adjusted for what everyone else was doing, it ranks fourth on the list above rather than first, and that is not a demotion. It is just the correct comparison.
Greg Maddux, 1994 and 1995
Maddux is the only man with two seasons in the top five, and they were consecutive.
1994: a 1.56 earned run average in a league averaging 4.22. 1995: a 1.63 in a league averaging 4.18. He won the Cy Young award in both, and in 1993 and 1992 as well, which is four in a row.
What makes them stranger than Martinez's is how he did it. Maddux struck out 156 men in 1994 and 181 in 1995, which are ordinary totals. He was not overpowering anybody. He gave up nothing because hitters kept putting the ball in play weakly and the ball kept going to somebody.
1994 was also the strike season, so his 202 innings came in a schedule cut short in August. He was 28.
Jacob deGrom, 2018, and 10 wins
deGrom had a 1.70 earned run average across 217 innings in 2018, with 269 strikeouts, in a league averaging 4.03.
His record was 10 and 9.
Across his 32 starts the New York Mets scored 3.53 runs a game for him. That is not a catastrophe, it is simply below average, and it was enough to turn the best pitching season of the decade into a 10 and 9 record.
He won the Cy Young award anyway, which would not have happened twenty years earlier, and that vote is probably the clearest moment the sport publicly stopped caring about pitcher wins. Voters looked at a 10 and 9 record and a 1.70 earned run average and gave the award to the second number.
He won it again in 2019.
The strikeout seasons
A different question, which the era-adjusted list does not answer: who was hardest to make contact against?
| Pitcher | Season | K per 9 | SO | BB | |---|---|---|---|---| | Gerrit Cole | 2019 | 13.82 | 326 | 48 | | Spencer Strider | 2023 | 13.55 | 281 | 58 | | Randy Johnson | 2001 | 13.41 | 372 | 71 | | Pedro Martinez | 1999 | 13.20 | 313 | 37 | | Chris Sale | 2017 | 12.93 | 308 | 43 | | Randy Johnson | 2000 | 12.56 | 347 | 76 | | José Fernández | 2016 | 12.49 | 253 | 55 | | Gerrit Cole | 2018 | 12.40 | 276 | 64 |
Every season on that list came after 1998, which is the same story the career strikeout list tells: the modern game misses bats at a rate no earlier era approached.
Randy Johnson's 2001 is the only one that combines the rate with real volume, 372 strikeouts across 250 innings. And Pedro's 1999 stands out for the other column: 313 strikeouts against 37 walks is a ratio nobody else on the list is near.
And the relievers
Every list above is starters only, because the innings qualifier excludes anyone who did not pitch one inning per team game. Relief seasons need their own table, and the numbers get absurd:
| Pitcher | Season | ERA | Index | IP | SV | |---|---|---|---|---|---| | Zack Britton | 2016 | 0.54 | 783 | 67.0 | 47 | | Fernando Rodney | 2012 | 0.60 | 678 | 74.7 | 48 | | Emmanuel Clase | 2024 | 0.61 | 664 | 74.3 | 47 | | Dennis Eckersley | 1990 | 0.61 | 638 | 73.3 | 48 | | Blake Treinen | 2018 | 0.78 | 544 | 80.3 | 38 |
An index of 783 means Britton allowed roughly an eighth of the runs an average pitcher would have in the same innings. He saved 47 games and gave up four earned runs all season.
These belong in a separate conversation rather than the same one, because 67 innings is a quarter of what Gibson threw and a reliever picks his moments in a way a starter cannot. But Eckersley's 1990 deserves its own line regardless of category: 73 and a third innings, 48 saves, 73 strikeouts, and four walks. Not four intentional walks. Four.
Fernando Rodney appearing here is also the best single argument against judging a career by its best year. He posted the second-greatest relief season on record in 2012 and finished his career having pitched for eleven different clubs, which puts him on a rather different list.
So what was the best season
The honest answer is that Pedro Martinez in 2000 and Dutch Leonard in 1914 are separated by one point on a measure precise to nowhere near one point, and picking between them is a preference rather than a finding.
But there is a tiebreak worth applying. Leonard's 0.96 came in a league where the average pitcher allowed 2.73, which means the whole environment was suppressing offence and he was the best of a suppressed group. Martinez's 1.74 came in a league where the average pitcher allowed 4.92, in the highest-scoring era since 1930, against lineups built entirely around power.
Being three runs better than everybody when everybody is giving up five is a harder thing than being two runs better when everybody is giving up three. The index treats those as equivalent. They are not quite.
Pedro Martinez, 2000. And if you want the raw number, Gibson's 1.12, which will also never be beaten, for a completely different set of reasons.
Knowing which season belongs to which pitcher is most of a trivia round. Test it against the clock in Pitch Clock, or pick the true line out of three in Two Truths.