To test the hypothesis that phasic muscle activation is related to the acceleration-deceleration characteristics of the resulting movement, we examined the relation between the duration of the acceleratory phase of a variety of movement types and the duration of the phasic muscle activity producing the acceleration (the initial agonist burst, AG1). Movements of five types were studied: (1) step-tracking movements of different amplitudes (10-90 deg) and durations (200-800 ms), (2) movements of the same amplitude (40 deg) and duration (600 ms) varying only in their symmetry ratio (SR, ratio of acceleration to deceleration durations), (3) movements in which acceleration duration was changed while acceleration magnitude was held constant, (4) oscillatory movements of different frequencies and peak amplitudes, (5) step-tracking movements against different inertial loads. Subjects made movements about the elbow joint in the horizontal plane. Surface electromyographic (EMG) activity was recorded from the biceps and the lateral head of the triceps muscles. Under all movement conditions tested and with acceleration duration ranging from 100 to 500 ms, acceleration duration varied linearly with the duration of AG1. Correlation coefficients for the linear regression lines ranged from 0.8 to 0.99. The slope of the best fit linear regression lines ranged from 0.5 to 1.6 and tended to be higher for extensions than flexions. The variations in slope may arise from differing mechanical properties of the biceps and triceps muscles, as well as from active forces produced in the antagonist. AG1 duration was unchanged by inertial loading when subjects kept acceleration duration constant. If subjects responded to an increase in inertial load with an increase in acceleration duration, there was a corresponding increase in AG1 duration. The data demonstrate a general relation between one characteristic of muscle activation (AG1 duration) and the resulting movement. The linear form of the relation is invariant across movement amplitude (range 10-90 deg), speed, duration (range 200-800 ms) and temporal profile (SR range 0.3-2.7), and is also independent of movement type (step, oscillatory). Such a general and simple relation between EMG and movement suggests that, at least to a first approximation, the nervous system can rather simply determine the muscle activation patterns needed to produce movements with desired characteristics.