The dorsal premotor area (PMd) of monkeys has been implicated in processes relating to movement preparation and movement selection. In the present study, we sought to determine whether PMd neurons that are activated during a delayed reaching task have directional responses that reflect either the target (i.e., the goal) of an intended movement or the physical properties of the movement itself. Two macaque monkeys were trained to perform a visually instructed, delayed reaching task with indirect visual feedback. The subjects and methods were identical to those described in the preceding paper. In the behavioral task, each subject moved a two-dimensional joystick with the right forelimb to align a cursor with targets presented on a video display. The paradigm dissociated the direction of forelimb movement from the spatial location of the target. This was accomplished by varying the spatial mappings between joystick and cursor. A variable delay separated the visual stimulus that instructed the target location (IS) from the visual stimulus that triggered the instructed movement (TS). Task-related activity was recorded from a total of 181 PMd neurons. The focus of this study was on directionally tuned neuronal responses that included 1) stimulus-related activity (phasic, following IS); 2) set-related activity (phasic, following IS and TS); and 3) movement-related activity (phasic, following TS). Of the entire sample of PMd neurons with directionally tuned activity, 114 were tested with two joystick/cursor mappings, permitting dissociation of directional responses that depended on limb trajectory from those that depended on target location. Task-related neuronal activity was classified as target-dependent if it covaried exclusively with target location across both conditions, and as limb-dependent if it covaried exclusively with limb trajectory. Directional activity that changed significantly across rotation conditions was classified as complex. Approximately one half of the sample of PMd neurons showed stimulus-related activity that was directionally tuned (56%, 64 of 114). Nearly all of the directionally classifiable stimulus-related activity was target dependent (94%, 44 of 47 responses), and none was limb dependent. A small proportion was classified as complex (6%, 3 of 47 responses). More than two thirds of the PMd neurons showed set-related activity that was directionally tuned (69%, 79 of 114). Among cells with set-related activity that was directionally classifiable, there were approximately 9 times as many target-dependent responses (76%, 48 of 63) as there were limb-dependent responses (8%, 5 of 63), with the remainder being complex (16%, 10 of 63). Approximately three quarters of the sample of PMd neurons showed early movement-related activity (before movement onset) that was directionally tuned (78%, 89 of 114). Among those cells whose early movement-related activity was directionally classifiable, there were > 3 times as many target-dependent responses ((51%, 34 of 66) as limb-dependent responses (14%, 9 of 66), with the remainder being complex (35%, 23 of 66). Approximately two thirds of the sample showed late movement-related responses (after movement onset) that were directionally tuned (68%, 78 of 114). Among those cells whose late movement-related activity was directionally classifiable, there were comparable numbers of target-dependent (25%, 15 of 61) and limb-dependent responses (28%, 17 of 61), with the remainder being complex (47%, 29 of 61). These results indicate a preferential representation of target location rather than limb trajectory among PMd neurons. Over the extended interval from IS to motor response, there was a gradual decline in the frequency of target-dependent activity and corresponding increases in the frequencies of both limb-dependent and complex activity. These findings suggest that PMd neurons may participate in mediating the sensory-to-motor transformation required by the delayed reaching