This study examined the effects of caffeine (CAF) supplementation on game-related statistics and perceptual responses of male basketball players during official games. Eight players (23.5±5.7 years) participated in this double-blind, placebo-controlled, randomized crossover study. Sixty minutes before the start of each game, players ingested a dose of either CAF (3 mg/kg body mass) or placebo (PLA) in a randomized order. Game-related statistics were collected. After the games, players reported ratings of perceived exertion (RPE) and their perceptions of endurance and power. Separate linear mixed models evaluated the effects of treatment (CAF, PLA), game location (home, away), and opponent level (better, worse) on game-related statistics and perceptual responses. Players committed fewer fouls when taking CAF compared to PLA (P<0.001, ES=0.75, moderate). No effects of CAF were found for the other game-related statistics (all P>0.05). RPE was higher in CAF compared to PLA (P=0.031, ES: 0.49, small). There were no differences in perceptions of endurance (P=0.388) or power (P=0.988). No effects of game location or opponent level were found for game-related statistics or perceptual responses (all P>0.05). This study suggests that caffeine supplementation can improve defensive performance in semiprofessional basketball players while also possibly increasing physical performances, seen the higher RPE.