The purpose of this study was to determine the reproducibility of time at or above 90 % (t (90 % )VO (2max)) and 95 % (t (95 % )VO (2max)) maximal oxygen uptake during an intermittent treadmill run to exhaustion. Twenty-two distance runners (age 38.0 +/- 7.1 yrs) performed two identical incremental and two identical intermittent tests on four separate days. Respiratory exchange was measured continuously throughout each test by an automated open-circuit gas analysis system. The incremental test consisted of increases in treadmill speed every minute until volitional exhaustion. The highest averaged 30-s oxygen uptake (VO (2)) value was defined as VO (2max) and the minimum speed that elicited VO (2max) was defined as vVO (2max). The intermittent test consisted of 30-s work intervals ran at 105 % vVO (2max) interspersed by 30-s relief intervals ran at 60 % vVO (2max) and was continued until volitional exhaustion. The time that VO (2) was at or above 90 % and 95 % of the mean maximum values elicited during the two previous incremental tests was determined for the intermittent tests. The mean t (95 % )VO (2max) was 232 (SD 174) s and 244 (SD 195) s and the mean t (90 % )VO (2max) was 480 (SD 220) s and 488 (SD 252) s, for trial 1 and trial 2, respectively. Reproducibility statistics for t (95 % )VO (2max) and t (90 % )VO (2max), respectively, were: 95 % limits of agreement 12 +/- 227 s and 8 +/- 328 s; coefficient of variation 34.5 % and 24.5 %; and intraclass correlation coefficient 0.80 and 0.75. Statistical power analysis indicated that this level of reproducibility would allow mean differences of 15 - 20 % between intermittent training protocols to attain statistical significance in future experimental research, with sample sizes probably within the resources of most researchers.