Using an exercise device that integrates maximal voluntary static contraction (MVC) of knee extensor muscles with dynamic knee extension, we compared progressive muscle fatigue, i.e., rate of decline in force-generating capacity, in normoxia (758 Torr) and hypobaric hypoxia (464 Torr). Eight healthy men performed exhaustive constant work rate knee extension (21 +/- 3 W, 79 +/- 2 and 87 +/- 2% of 1-leg knee extension O2 peak uptake for normoxia and hypobaria, respectively) from knee angles of 90-150 degrees at a rate of 1 Hz. MVC (90 degrees knee angle) was performed before dynamic exercise and during <or = 5-s pauses every 2 min of dynamic exercise. MVC force was 578 +/- 29 N in normoxia and 569 +/- 29 N in hypobaria before exercise and fell, at exhaustion, to similar levels (265 +/- 10 and 284 +/- 20 N for normoxia and hypobaria, respectively; P > 0.05) that were higher (P <0.01) than peak force of constant work rate knee extension (98 +/- 10 N, 18 +/- 3% of MVC). Time to exhaustion was 56% shorter for hypobaria than for normoxia (19 +/- 5 vs. 43 +/- 7 min, respectively; P <0.01), and rate of right leg MVC fall was nearly twofold greater for hypobaria than for normoxia (mean slope = -22.3 vs. -11.9 N/min, respectively; P <0.05). With increasing duration of dynamic exercise for normoxia and hypobaria, integrated electromyographic activity during MVC fell progressively with MVC force, implying attenuated maximal muscle excitation. Exhaustion, per se, was postulated to related more closely to impaired shortening velocity than to failure of force-generating capacity.