The use of simple force feedback in an isometric muscle control system utilizing orderly recruitment of motor units is explored. Cat medial gastrocnemius motor units were stimulated with and without simple force feedback gain ranging from 0.7 to 0.9. Ramp, triangular, staircase, sinusoidal and bandwidth-limited pseudo-random input recruitment signals were used to study tracking accuracy through linear correlation in ramp and triangular signals, cross correlation in sinusoidal and random signals, and rise time and steady state error in staircase signals. Dramatic improvements were found in most tested tracking variables with the use of feedback; squared correlation coefficients increased from a mean of 0.93 to 0.99 for ramp signals and from 0.76 to 0.98 in triangular signals. Mean peak cross-correlations improved from 0.85 to 0.98 in random signals and from 0.93 to 0.98 for sinusoidal inputs, and mean time to peak cross-correlations decreased from 144 to 24 ms in random signals and from 156 to 25 ms in sine waves. Rise times in staircase signals decreased from a mean of 520 to 175 ms, and mean steady state error decreased from 12 to 3%. Significant effects of the triangle cycle time, sinusoidal frequency and staircase step were found on the performance of the muscle force control system. In addition, the possible effects of intrinsic feedback mechanisms on the control system were examined by repeating the closed loop part of the study but with the sciatic nerve cut proximally. The tracking results were essentially and statistically the same as in the closed loop condition. It was concluded that a simple feedback configuration provided superior tracking performance for a practical application in which orderly recruitment is used to control muscles; furthermore, it was concluded that this type of system would be virtually immune to external disturbances such as spasticity resulting from intact spinal neural feedback mechanisms found in paralyzed individuals.