Bicuculline (10–30 μg, but usually 30 μg) was injected locally into 20 different sites in the dorsolateral prefrontal cortex (PFC) of 2 Japanese macaque monkeys, while they were performing a delayed response task. The task was initiated by the rotation of a handle to a central zone by the wrist joint and consisted of seven periods: an initial waiting period of 0.3 s, a pre-cue period (central green lamp of 1.0 s), a cue period (left or right green cue of 0.3 s), a delay period of 4.0 s (occasionally 1 s), a go period (central red lamp; rotation of the handle to either the left or right zone within 1.0 s), a hold period (holding of the handle in either the left or the right zone), and a final reward period. The parameters of the task performance, such as the frequency of correct trials, the frequency of directional error trials in which the monkeys rotated the handle in an incorrect direction during the go period, and the frequency of omission error trials, in which the monkeys did not rotate the handle during the go period, were examined before and after the injection of bicuculline. The injections of bicuculline induced a burst of multi-neuronal activity around the sites of injection. Within 5 min of an injection into one of 7 different sites in the PFC, three different kinds of performance deficit were observed: 1) an increase in the frequency of error responses during the go period in both left-cue and right-cue trials, after injection into the dorso-caudal portion of the principal sulcus (2 sites); 2) an increase in the frequency of directional error responses during the go period in either left-cue or right-cue trials, after injection into the bottom of the middle principal sulcus (3 sites), and 3) an increase in the frequency of omission of responses during the go period, after injection into the dorsal region of the caudal principal sulcus (2 sites). Injections at the remaining 13 sites did not induce any deficits, although injections into the dorsal bank of the principal sulcus (3 sites) induced a decrease in the frequency of the task trials as a result of prolonged intertrial intervals (ITIs). Our results suggest that locally disturbed neuronal activity in different small areas of the PFC induces different deficits in the performance of the delayed response task. Neuronal activity in different, localized areas of the PFC may be involved in different processes of a performance that is based on spatial short-term memory.