Many team sports require participants to repeatedly produce maximal or near maximal sprints of short duration (1-7s) with brief recovery periods, over an extended period of time (60-90 min). Therefore, an important fitness component for these sports is what is often termed repeated sprint ability (RSA). A variety of tests have been devised to assess RSA. However, while the reliability of some of these tests has been reported, there has been no published attempt to assess the validity of RSA tests. The purpose of the present paper was to assess the validity of a popular RSA test--the 5 x 6 s cycle test. Ten moderately-trained males (Mean+/-SD age: 23.6+/-3.0 yrs, mass: 71.8+/-7.4 kg, peak VO2: 46.32+/-4.16 ml.kg(-1)min(-1)) participated in the investigation. Following familiarisation, subjects reported to the laboratory at the same time of day, on three occasions, each separated by at least 48 hours. Testing consisted of a 5 x 6 s cycle test (5 x 6 s maximal sprints every 30 s), a graded exercise test (GXT) and a simulated game (3 x 15 min periods separated by five min). The simulated game consisted of a 1-min circuit that was repeated 15 times in each period. The circuit replicated typical movement patterns observed during motion analysis of field hockey games. Each circuit commenced with a 15-m maximum sprint through timing gates. Time to run 5, 10 and 15 m was recorded. Oxygen consumption was measured during both the GXT and the simulated game with a portable gas analysis system (Cosmed K4 b2, Italy). Correlation coefficients between dependent variables were calculated using Pearson's Product Moment (r). There was a significant correlation between power decrement during the 5 x 6 s cycle test and decrement in 15-m time across the three periods (r = 0.76, P<0.05), but not decrement in 10-m time (r = 0.54) or 5-m time (r = 0.42). These results suggest that the 5 x 6 s cycle test is valid for assessing the decrement in 15-m time, but not the decrement in 5 or 10m time. Thus, one measure of RSA (sprint decrement) appears to be specific to the test protocol, rather than a general quality. The most likely explanation is that the energy requirements of the 5 x 6 s cycle test more closely match those required to repeatedly run 15 m (mean time = 2.74 s) than to repeatedly run 10 m (mean time = 1.97 s) or 5 m (mean time = 1.13 s). It is therefore suggested that, while the 5 x 6 s cycle test is often used to assess RSA ability in a wide range of sports, it may need to be modified to reflect the common sprint distances found in specific sports.