This study used a novel feedback approach to control a robotic foot using force and length signals transmitted from an isolated Xenopus laevis frog muscle. The foot's environment (inertial versus hydrodynamic), gearing (outlever/inlever) and size were changed to alter the muscle's load. Upon nerve stimulation (250 Hz, 80 ms train duration), variation in loading generated a range of muscle stress (19.8±5.3 to 66.0±22.5 kPa), work (1.89±0.67 to 6.87±2.96 J kg(-1) muscle) and power (12.4±7.5 to 64.8±28.3 W kg(-1) muscle; mean ± s.d., N=6 frogs). Inertial versus hydrodynamic loading dramatically shifted contractile dynamics. With the foot in water, the muscle generated ∼30% higher force, yet shortened slower, producing lower power than inertial loading. Power increased in air from 22.6±5.8 to 63.6±27.2 W kg(-1) muscle in response to doubling the gear ratio, but did not increase in water. Surprisingly, altering foot size diminished muscle performance in water, causing power to drop significantly from 41.6±11.1 to 25.1±8.0 W kg(-1) muscle as foot area was doubled. Thus, morphological modifications influenced muscle dynamics independently of neural control; however, changes in loading environment and gearing affected contractile output more strongly than changes in foot size. Confirming recent theory, these findings demonstrate how muscle contractile output can be modulated solely by altering the mechanical environment.