Previous studies suggest that stimulation of lateral intraparietal cortex (LIP) evokes saccadic eye movements towards eye-fixed or head-fixed goals, whereas most single-unit studies suggest that LIP uses an eye-fixed frame with eye position modulations. The goal of our study was to determine the reference frame for gaze shifts evoked during LIP stimulation in head-unrestrained monkeys. Two macaques (M1 and M2) were implanted with recording chambers over the right intraparietal sulcus and with search coils for recording 3-D eye and head movements. The LIP region was microstimulated using pulse trains of 300Hz, 100-150µA, and 200ms. 85 putative LIP sites in M1 and 194 putative sites in M2 were used in our quantitative analysis. Average amplitude of the stimulation-evoked gaze shifts was 8.67° for M1 and 7.97° for M2, with very small head movements. When these gaze shift trajectories were rotated into three coordinate frames (eye, head, and body), gaze end-points distribution for all sites was most convergent to a common point when plotted in eye coordinates. Across all sites, the eye-centered model provided a significantly better fit compared to the head, body, or fixed-vector models (the latter model signifies no modulation of the gaze trajectory as a function of initial gaze position). Moreover, the probability of evoking a gaze shift from any one particular position was modulated by the current gaze direction (independent of saccade direction). These results provide causal evidence that motor commands from LIP encode gaze command in eye-fixed coordinates but are also subtly modulated by initial gaze position.