Twenty-three male college athletes performed submaximal repetition tests (70, 75, 80, 85, 90, and 95% 1-RM) in the bench press (BP), squat (S), and power clean (PC) lifts. For each lift the best predictor of 1-RM strength was defined as the maximal number of repetitions performed at a given lifting intensity (i.e., %1-RM) which represented the highest prediction coefficient (multiple R). ANOVA revealed that regardless of lift, the number of repetitions significantly decreased (p < 0.05) as lifting intensity increased. The best predictor for BP was the number of repetitions performed at 95% 1-RM. For S and PC lifts the best predictors corresponded to the number of repetitions at 80 and 90%, respectively. The S best predictor had the highest prediction power (R2 accounted for 26.9% of the variance). The BP and PC best predictors accounted for 11.6 and 19.1% of the variance, respectively. Although the corresponding best predictors (multiple R) for each lift represented different percentages of 1-RM, their respective predictive power (R2) was not significantly different (p > 0.05).