This study compared between performance indices of repeated sprint tests (RSTs) (12 x 20 m) during different stages of basketball game. Twelve young (17 +/- 0.5 year) basketball players performed three RSTs (after a warm-up, at half-time and after a full game) and aerobic power test, each on different days. Ideal (fastest) sprint time (IS) was significantly faster at half-time (p < 0.007) compared to after warm-up. There was no difference between IS after the warm-up and after a full game. Total (accumulative) sprint time (TS) was significantly faster at half-time (p < 0.03) compared to after the warm-up. There was no difference between TS after the warm-up and after a full game. No differences were found in the performance decrement (PD) between the three RSTs. Significant negative correlations were found between predicted VO(2) max and PD during the 12 x 20 m RST only when the RST was performed at half-time (r = -0.58) and after a full game (r = -0.59), and not when the RST was performed after the warm-up. The findings suggest that a more intense warm-up is needed for better repeated sprint performance at the initial phases of the game and that the aerobic system is important to intensity maintenance mainly during the last stages of the game.