Results-One fifth of players did not attend any rugby training during the 16 week summer period; the remainder attended a median of 14 sessions. Throughout the 1997-1998 season, 675 injury episodes occurred to 423 (53%) players during training or in matches. After adjustment for whether players held a professional contract or were amateurs, Cox regression showed a 3.9% relative increase (95% confidence interval (CI) 1.9 to 5.9%) in the risk of injury over the season for each additional preseason training week attended, and a 61% relative increase (95% CI 32 to 97%) for those players who had been injured or were carrying an injury at the end of the previous season. Conclusions-Injury risk is more likely to be related to rugby training (type of activities undertaken in rugby training, or personalities and characteristics of players undertaking training more frequently) than to overall player fitness. Players who were injured at the end of the previous season were more likely to be injured in the following season. This may be because they do not allow previous injuries to heal suYciently before returning to the game, or the intensity of their participation may increase their risk of injury. (Br J Sports Med 2001;35:412-417)