Muscle samples were taken from the gluteus, semitendinosus and longissimus muscles of a captive cheetah immediately after euthanasia. Fibres were "skinned" to remove all membranes leaving the contractile filament array intact and functional. Segments of skinned fibres from these cheetah muscles and from rabbit psoas muscle were activated at 20°C by a temperature jump protocol. Step and ramp length changes were imposed after active stress had developed. The stiffness of the non-contractile ends of the fibres (series elastic component) was measured at two different stress values in each fibre; stiffness was strongly dependent on stress. Using these stiffness values, the speed of shortening of the contractile component was evaluated, and hence the power it was producing. Fibres were analysed for myosin heavy chain content using gel electrophoresis, and identified as either slow (Type I) or fast (Type II). The power output of cheetah Type II fibre segments was 92.5 ± 4.3 W kg-1 (mean ±s.e., 14 fibres) during shortening at relative stress 0.15 (=stress during shortening/isometric stress). For rabbit psoas fibre segments (presumably Type IIX) the corresponding value was significantly higher (P<0.001), 119.7 ± 6.2 W kg-1 (mean ±s.e.,7 fibres). These values are our best estimates of the maximum power output under the conditions used here. Thus the contractile filament power from cheetah was less than that of rabbit when maximally activated at 20°C, and does not account for the superior locomotor performance of the cheetah.