Isolated rat diaphragm muscle was stimulated repetitively to induce fatigue, and the work done during each contraction was measured. Work per cycle was calculated by measuring force as the activated muscle was subjected to sinusoidal length changes (from 97 to 103% of L0, where L0 is rest length). Work was calculated from the loop formed when force was plotted against length. Work done was positive when the muscle was shortening and was negative when it was lengthening; net work was the difference. Work output was varied by changing the stimulus duty cycle (4, 8, or 16% of the total cycle duration) and cycle frequency (1, 2, or 4 Hz). The rate and extent of the decrease in power was influenced much more by changes in cycle frequency than by changes in duty cycle. Duty cycle and cycle frequency combinations that resulted in greater power in the prefatigue trials were associated with a more rapid rate of fatigue. However, net positive power at the end of the 15-min fatigue period was greater under these same conditions (i.e., high duty cycle and high cycle frequency). Fatigue in working diaphragm muscle depends more on cycle frequency than on duty cycle.