The purpose of this study was to determine whether maximal oxygen uptake (VO2max) is protocol dependent during arm cycle ergometry (ACE) for quadriplegic males with spinal cord injuries (SCI). Twenty-four non-ambulatory subjects (aged 20-38 yr) with cervical SCI were divided into two groups based on wheelchair sports classification (IA group = 14; IB/IC group = 10). They underwent three different, continuous graded exercise tests spaced at least 1 wk apart on an electronically braked arm cycle ergometer. Following a 3-min, unloaded warm-up at 60 rpm, the work rate was increased 2, 4, or 6 W.min-1 for the IA group and 4, 6, or 8 W.min-1 for the IB/IC group. Ventilation and gas exchange were measured breath-by-breath with a SensorMedics 4400 computerized system. Repeated-measures ANOVA showed no significant difference among the three protocols for VO2max in the IA group (P greater than 0.05). The mean (+/- SD) VO2max values (ml.kg-1.min-1) were 10.8 (+/- 3.4), 11.0 (+/- 2.7), and 10.2 (+/- 2.9) for the 2, 4, and 6 W.min-1 protocols, respectively. In contrast, the IB/IC group showed a significant difference among the protocols for VO2max (P less than 0.05). The mean (+/- SD) VO2max values (ml.kg-1.min-1) were 16.8 (+/- 4.5), 15.3 (+/- 4.3), and 14.6 (+/- 4.3) for 4, 6, and 8 W.min-1, respectively. Post hoc analysis revealed a difference between the 4 and 8 W.min-1 protocols. Our results suggest that graded exercise testing of SCI persons with quadriplegia, using ACE, should employ work rate increments between 2-6 W.min-1 and that work rate increments of 8 W.min-1 or greater will underestimate VO2max.