The purpose of this study was to determine (1) the repeatability of a currently recommended protocol for determination of VO2 peak during upper body exercise involving a crank rate of 60 rev min-1 (Hale et al. 1988), (2) whether employing a higher crank rate (70 rev min- 1) elicits higher VO2 peak values; (3) whether including an additional exercise stage 5 min after volitional exhaustion would elicit higher VO2 peak values. Twenty subjects (mean ± SD) age, height and body mass were 22·8 ± 1·8 years, 176 ± 5·0 cm and 77·5 ± 4·0 kg, respectively) visited the laboratory on three separate occassions. Subjects performed a different test to determine upper body VO2 peak on each visit. The tests were continuous and incremental in nature and were performed on a Monark cycle ergometer (Model 814E) adapted for arm exercise. Test 1 required subjects to exercise at a crank rate of 60 rev min-1 (60RPM-1). Test 2 was identical to test 1 (60RPM-2) but after 5 min recovery from the test subjects performed a further exercise stage (60RPM-F). Test 3 employed a crank rate of 70 rev min- 1 (70RPM). Five minutes after volitional exhaustion a further exercise stage was performed (70RPM-F). The levels of agreement for the 60 rev min- 1 protocol were ± 0.50 1 min- 1. All VO2 peak values achieved were within these limits. However, a higher VO2 peak was achieved by employing a crank rate of 70 rev min- 1 (p< 0.05). The VO2 peak achieved for 60RPM-F was similar to 60RPM whereas the VO2 peak for 70RPM-F tended to be less than that for 70RPM (p = 0.064). The results of this study suggest that the 60 rev min- 1 protocol can reliably determine VO2 peak, however it would appear to be more appropriate to employ a 70 rev min- 1 protocol as higher values were attained.