Are NBA injuries
really getting worse?
Yes. Rotation players missed 16.8% of their teams' games from 1990 to 2010 and 27.0% in 2025-26, the most in 37 seasons. But it's not the injuries you'd guess: most of the rise is absences of 1 to 3 weeks, and stars went from the most available players in the league to the least.
Share of games rotation players missed, 1990-2010 vs 2025-26. A typical rotation player used to miss about 1 game in 6. Now it's more than 1 in 4, the highest of all 37 seasons.
From 1990 to 2010, stars missed fewer games than other rotation players (13.8% vs 17.1%). Over the last 4 seasons they missed more (26.9% vs 24.7%). That's about 11 games a season per star then, 22 now.
Of the rise since 1990-2010, about 2 points are single-game absences (where rest days live) and just 1 is absences of 20+ games. The other 5 come from absences of 2 to 19 games: 1 to 3 weeks.
Watch the model get built, version by version.
Share of games missed
Rotation players, all absences, 1989-90 to 2025-26. The jump starts in 2017-18, before COVID, and each of the last 4 seasons is higher than any season before 2018. Hover or tap a bar.
The usual explanations don't fit
The 4 explanations I hear most, tested against the data.
Faster games
Teams play faster than in the 2000s (91.5 possessions), but the early 1990s (96.9) were nearly as fast as today (99.0), and those were the healthiest seasons in my data.
Older players
Rotation players are younger now (26.5) than in 1990-2010 (27.6). Adjusting for age, minutes and star status barely moves the rise (5.9 points raw, 5.6 adjusted).
More back-to-backs
The schedule got easier: about 21 back-to-backs a team per season in 1990-2010, about 14 now.
Heavier minutes
Rotation players average 27 minutes a game now, against 28.5 back then. They play less, rest more, and still miss more games.
What does fit is the shape of the absences: almost twice as many injury-type absences per game played (22.4 → 40.3 per 1,000 games), but each one a little shorter (7.7 → 6.8 games). That points to teams managing bodies differently, although box scores can't prove it. And a 2024 report the NBA commissioned found that resting players did not lower their risk of getting hurt later in the season: rest didn't buy health.
Severe injuries and the Achilles
Short absences are up a lot. Severe ones are up a little.
Per 100 rotation players a season, 1990-2010 vs the last 4 seasons: a rise of about 25%, much smaller than everything else. By games lost to long absences, 2025-26 ranks just 5th, behind seasons like 1996-97.
7 players tore an Achilles during the 2024-25 season or playoffs: James Wiseman (IND), Isaiah Jackson (IND), Dru Smith (MIA), Dejounte Murray (NOP), Damian Lillard (MIL, playoffs), Jayson Tatum (BOS, playoffs), Tyrese Haliburton (IND, playoffs). At the 1990-2024 rate that happens about once in 10,000 seasons, so it was a real cluster. But 2025-26 had only 1, and without that one record year the long-term rise isn't statistically significant. ACL tears have stayed flat at about 2 a season.
Before the 2026-27 opener there were already 3 more full Achilles tears in the preseason: Ariel Hukporti (PHI), Otega Oweh (OKC), Ben Simmons (SAC).
2026-27, as of Oct 27, 2026
A snapshot from the video, not a live tracker: each team's first 3 games (Oct 20-27, 2026). The big injury headlines came in the days right after it.
of games missed (108 of 696 player-games). That ranks 16th of 38 seasons at the same point, lower than 20.4% a year earlier.
of games missed by last season's All-Stars and All-NBA players. That ranks 25th of 38.
In past seasons the opening week only loosely predicted the full season. The opening week was normal; the full season is still open.
All 37 seasons
Share of their teams' games that rotation players missed, split by how long each absence lasted. Tap a column to sort. Stars = All-Star or All-NBA the season before. Achilles = full tears during the season or playoffs.
How the model works
V1 · The raw count
Total games rotation players missed each season: about 3,363 a season in 1990-2010, 6,704 in 2025-26. Unfair on its own: the league added 3 teams and coaches spread minutes across more players.
V2 · A fair rate
Each player's games missed divided by the games his team played while he was on the roster, then split by absence length: 1 game, 2 to 19, and 20+.
V3 · Who, and why
Stars vs everyone else, then the usual explanations: pace, age, back-to-backs and minutes, with an age-adjusted check.
V4 · Severe injuries
Absences of 20+ games, plus a hand-checked list of full Achilles and ACL tears during the season or playoffs.
V5 · Noise and a backtest
10,000 simulated seasons where nothing changed after 2010, each built from random real 1990-2010 player seasons. They landed between 14.55% and 19.53%, and none reached even the lowest of the last 4 real seasons (22.8%, 2023-24). A backtest predicts each past season using only the seasons before it: the "nothing changes" forecast was too low in 26 of 27 seasons since 1999-2000, off by about 5 points a year since 2018.
What counts as a game missed
Injury reports keep changing (injured lists, official reports, resting rules, the 65-game minimum), so I used the one thing that hasn't changed since the 1980s: the box score. Either you played or you didn't.
955,424 regular-season box-score lines, 1986-87 to Oct 27, 2026. A rotation player is anyone who plays about 20 minutes a night. If he isn't in his team's box score, that's a game missed.
A box score can't say why a player sat: injury, illness, rest or suspension all count. That's why later versions split absences by length. 2019-20 and 2020-21 were shortened seasons, and 2021-22 includes COVID protocol absences.