From SABR member J.G. Preston at The J.G. Preston Experience on February 22, 2015:
I wish I could cite the exact source of this, but I believe it was Bill James who speculated the most important factor in causing a starting pitcher to tire may not be the number of innings pitched or the number of pitches thrown, but the length of time he works. If that’s true, it could be that one of the reasons we have seen such a steep decline in complete games in recent years — not the primary reason, maybe not even one of the biggest reasons, but a contributing factor — is that it’s taking longer to play nine innings these days.
With that in mind, I went to Baseball-Reference.com’s Play Index to look for the longest complete game performances by length of game, in time (as opposed to innings). A few caveats here. First of all, there are plenty of errors in the time of game data. For instance, B-R.com shows the longest game in its database (at least as I write this) is a 1941 nine-inning game between the Red Sox and Yankees that for some reason (surely a typographical error) shows up as 348 hours and 55 minutes. The attendance is also listed as 143, so you can see some odd keyboarding was going on…in fact, it looks like the time and attendance were transposed. Retrosheet, where B-R.com gets is data, shows the time of game as 2:23 with attendance of 20,935. Sometimes the typos are on Retrosheet’s end; for instance, this 1952 nine-inning complete game by Lou Kretlow is listed as 4:47, but in contemporaneous newspaper box scores the time was listed as a more-sensible 2:47.
Originally published: February 25, 2015. Last Updated: February 25, 2015.