The ergogenic effects of caffeine in human exercise have been shown to improve endurance and anaerobic exercise performance. Previous work has demonstrated that 70 μM caffeine (physiological maximum) can directly increase mouse extensor digitorum longus (EDL) muscle power output (PO) in sprintlike activity by 3%. Our study used the work loop technique on isolated mouse muscles to investigate whether the direct effect of 70 μM caffeine on PO differed between 1) maximally and submaximally activated muscle; 2) relatively fast (EDL) and relatively slow (soleus) muscles; and 3) caffeine concentrations. Caffeine treatment of 70 μM resulted in significant improvements in PO in maximally and submaximally activated EDL and soleus (P < 0.03 in all cases). For EDL, the effects of caffeine were greatest when the lowest, submaximal stimulation frequency was used (P < 0.001). Caffeine treatments of 140, 70, and 50 μM resulted in significant improvements in acute PO for both maximally activated EDL (3%) and soleus (6%) (P < 0.023 in all cases); however, there was no significant difference in effect between these concentrations (P > 0.420 in all cases). Therefore, the ergogenic effects of caffeine on PO were higher in muscles with a slower fiber type (P < 0.001). Treatment with 35 μM caffeine failed to elicit any improvement in PO in either muscle (P > 0.72 in both cases). Caffeine concentrations below the physiological maximum can directly potentiate skeletal muscle PO. This caffeine-induced increase in force could provide similar benefit across a range of exercise intensities, with greater gains likely in activities powered by slower muscle fiber type.