We studied the activity of 123 cells in the arm area of the motor cortex of three rhesus monkeys while the animals performed a 2-dimensional (2-D) step-tracking task with or without a delay interposed between a directional cue and a movement triggering signal. Movements of equal amplitude were made in eight directions on a planar working surface, from a central point to targets located equidistantly on a circle. The appearance of the target served as the cue, and its dimming, after a variable period of time (0.5-3.2 s), as the "go" stimulus to trigger the movement to the target; in a separate task, the target light appeared dim and the monkey moved its hand towards it without waiting. Population histograms were constructed for each direction after the spike trains of single trials were aligned to the onset of the cue. A significant increase (3-4x) in the population activity was observed 80-120 ms following the cue onset; since the minimum delay was 500 ms and the average reaction time approximately 300 ms, this increase in population activity occurred at least 680-720 ms before the onset of movement. A directional analysis (Georgopoulos et al. 1983, 1984) of the changes in population activity revealed that the population vector during the delay period pointed in the direction of movement that was to be made later.