The relationship between force and electromyographic (EMG) signals of the cat soleus muscle was obtained for three animals during locomotion at five different speeds (154 steps), using implanted EMG electrodes and a force transducer. Experimentally obtained force-IEMG (= integrated EMG) relationships were compared with theoretically predicted instantaneous activation levels calculated by dividing the measured force by the predicted maximal force that the muscle could possibly generate as a function of its instantaneous contractile conditions. In addition, muscular forces were estimated from the corresponding EMG records exclusively using an adaptive filtering approach. Mean force-IEMG relationships were highly non-linear but similar in shape for different cats and different speeds of locomotion. The theoretically predicted activation-time plots typically showed two peaks, as did the IEMG-time plots. The first IEMG peak tended to be higher than the second one and it appeared to be associated with the initial priming of the muscle for force production at paw contact and the peak force observed early during the stance phase. The second IEMG peak appeared to be a burst of high muscle activation, which might have compensated for the levels of muscle length and shortening velocity that were suboptimal during the latter part of the stance phase. Although it was difficult to explain the soleus forces on the basis of the theoretically predicted instantaneous activation levels, it was straightforward to approximate these forces accurately from EMG data using an adaptive filtering approach.