1. In three human subjects, we measured the latency of pursuit and saccadic eye movements made to an eccentric target after a fixated central target was extinguished. In one set of experiments, we varied the time interval between the extinction of the central target and the appearance of the eccentric target ("gap duration"). In a second set of experiments, we varied the eccentricity at which the second target appeared. 2. Varying the gap duration produced similar changes in the latencies of pursuit and saccades. Gaps as short as 30 ms caused significant decreases in latency; progressively longer gaps produced shorter latencies, reaching a minimum for gaps of 150-200 ms. Over the range of gap durations used, the latencies of pursuit and saccades displayed the same dependence on gap duration. 3. Varying the eccentricity of the second target produced different effects on the latencies of pursuit and saccades. Saccade latencies increased when the eccentricity of the second target was decreased from 4 degrees to 0.5 degree, whereas pursuit latencies were not consistently altered. Despite these differences in the dependence on retinal eccentricity between pursuit and saccades, imposing a 200-ms gap between the extinction of the fixation point and appearance of the second target still reduced the latency of both. 4. Our results are consistent with the idea that the mechanisms underlying the release of fixation for pursuit and saccades have shared inputs but a different neural substrate. The common dependence on gap duration may indicate that a single preparatory input coordinates both types of movements. The different dependence on retinal eccentricity indicates that there are differences in the spatial organization of the premotor circuits that trigger the onset of the two types of movements.