We examined neuronal activity in three parts of the primate frontal cortex: the dorsal (PMd) and ventral (PMv) premotor cortex and a ventrolateral part of the dorsolateral prefrontal (PF) cortex. Two monkeys fixated a 0.2 degrees white square in the center of a video display while depressing a switch located between two touch pads. On each trial, a spatial-attentional/mnemonic (SAM) cue was presented first. The SAM cue consisted of one 2 degrees x 2 degrees square, usually red or green, and its location indicated where a conditional motor instruction would appear after a delay period. The stimulus event containing the motor instruction, termed the motor instructional/conditional (MIC) cue, could be of two general types. It might consist of a single 2 degrees x 2 degrees square stimulus identical to one of the SAM cues presented at the same location as the SAM cue on that trial. When the MIC cue was a single square, it instructed the monkey to move its forelimb to one of the two touch pads according to the following conditional rule: a green MIC cue meant that contact with the right touch pad would be rewarded on that trial and a red MIC cue instructed a movement to the left touch pad. Alternatively, the MIC cue might consist of two 2 degrees x 2 degrees squares, only one of which was at the SAM-cue location: in those cases, one square was red and the other was green. The colored square at the SAM cue location for that trial was the instructing stimulus, and the other part of the MIC cue was irrelevant. When, after a variable delay period, the MIC cue disappeared, the monkey had to touch the appropriate target within 1 s to receive a reward and could break visual fixation. The experimental design allowed comparison of frontal cortical activity when one stimulus, identical in retinocentric, craniocentric, and allocentric spatial location as well as all other stimulus parameters, had two different meanings for the animal's behavior. When a stimulus was the SAM cue, it led to either a reorientation of spatial attention to its location, or the storage of its location in spatial memory. By contrast, when it was the MIC cue, the same stimulus instructed a motor act to be executed after a delay period. For the majority of PMd neurons (55%), post-MIC cue activity exceeded post-SAM cue activity.(ABSTRACT TRUNCATED AT 400 WORDS)