Measurements were made from the human adductor pollicis muscle of force, contractile speed, and electromyographic activity (EMG) before, during, and after maximal isometric voluntary contractions sustained for 60 s. The use of brief test periods of maximal nerve stimulation with single shocks or trains of shocks enabled various muscle mechanical properties to be studied throughout each contraction. Electrical activity was measured after rectification and smoothing of the surface potentials and also by counting the total number of potentials per unit time from a population of motor units using fine wire intramuscular electrodes. During a 60-s maximal voluntary contraction, the force fell by 30-50%. Throughout the experiment the voluntary force matched that produced by supramaximal tetanic nerve stimulation. This indicated that, with sufficient practice, full muscle activation could be maintained by voluntary effort. However, the amplitude of the smoothed, rectifed EMG and the rate of spike counts declined. Since no evidence for neuromuscular block was found, the decline in EMG and spike counts was attributed to a progressive reduction of the neural drive from the central nervous system, despite maintained maximum effort. After the prolonged voluntary contractions twitch duration was prolonged, mainly as a result of slowing in relaxation rate. Twitch summation in unfused tetani increased. Both the maximum rate of relaxation and the time course of force decay declined by 50-70%. Similar changes were seen in both voluntary contractions and in test periods of stimulation. The percentage change in muscle contractile speed measured by these parameters approximately equaled the percentage change in the surface EMG measured simultaneously. It is concluded that 1) during a 60-s sustained maximal voluntary contraction there is a progressive slowing of contraction speed such that the excitation rate required to give maximal force generation is reduced, 2) the simultaneous decline in EMG may be due to a continuous reduction in motoneuron discharge rate, and 3) the EMG decline may not necessarily contribute to force loss.