Repeated sprint testing is gaining popularity in team sports, but the methods of data analysis and relationships to speed and endurance qualities are not well described. We compared three different methods for analyzing repeated sprint test results, and we quantified relationships between repeated sprints, short sprints, and endurance test scores. Well-trained male junior Australian Football players (n = 60, age 18.1 +/- 0.4 years, height 1.88 +/- 0.07 m, mass 82.0 +/- 8.1 kg; mean +/- SD) completed a 6 x 30-m repeated sprint running test on a 20-second cycle, a 20-m sprint test (short sprint), and the 20-m multistage shuttle run for endurance. Repeated sprint results were evaluated in three ways: total time for all six sprints (TOTAL), percent change from predicted times (PRED) from the fastest 30-m sprint time, and percent change from first to last sprint (CHANGE). We observed a very large decrement (CHANGE 6.3 +/- 0.7%, mean +/- 90% confidence limits) in 30-m performance from the first to last sprint (4.16 +/- 0.10 to 4.42 +/- 0.11 seconds, mean +/- SD). Results from TOTAL were highly correlated with 20-m sprint and 20-m multistage shuttle run tests. Performance decrements calculated by PRED were highly correlated with TOTAL (r = 0.91), but neither method was directly comparable with CHANGE (r = -0.23 and r = 0.12 respectively). TOTAL was moderately correlated with fastest 20-m sprint time (r = 0.66) but not the 20-m multistage shuttle run (r = -0.20). Evaluation of repeated sprint testing is sensitive to the method of data analysis employed. The total sprint time and indices of the relative decrement in performance are not directly interchangeable. Repeated sprint ability seems more related to short sprint qualities than endurance fitness.