We studied the role of the primate frontal eye fields (FEFs) in eye-head gaze shifts by recording EMG activity from multiple dorsal neck muscles after electrical stimulation of a broad distribution of sites throughout FEF. We assess our results in light of four mechanisms forwarded to account for why eye and head movements follow FEF stimulation. Two mechanisms propose that movements are generated indirectly by FEF stimulation in response to either a percept or an eccentric orbital position. Two other mechanisms propose that movements are evoked directly through the issuance of either a gaze command or separate eye and head commands. FEF stimulation evoked short-latency ( approximately 20 ms) neck EMG responses from the vast majority (>95%) of stimulation sites. Evoked responses usually preceded the gaze shift by approximately 20 ms, even for small gaze shifts (<10 degrees ) not typically associated with head motion. Evoked responses began earlier and attained a larger magnitude when accompanied by larger gaze shifts and took a form consistent with the recruitment of the appropriately directed head movements to accompany the evoked gaze shift. We also observed robust neck EMG even when stimulation failed to evoke a gaze shift and occasionally observed head-only movements when the head was unrestrained. These results resemble neck EMG evoked from the superior colliculus (SC). Neck EMG response latencies approached the minimal conduction time to the motor periphery and hence are not consistent with either of the indirect mechanisms. The widespread nature of the cephalomotor drive from the FEF, the scaling of neck EMG responses with gaze magnitude, and the consistently earlier generation of the EMG versus gaze response are difficult to reconcile with suggestions that separate FEF channels encode eye and head motion independently. The most parsimonious interpretation is that a gaze command issued by the FEF is decomposed into eye and head commands downstream of the SC. The relative timing of the neck EMG and gaze shift responses, and the presence of neck EMG responses on trials without gaze shifts, implies that head premotor elements are not subjected to the same brain stem control mechanisms governing gaze shifts.