Purpose: To investigate the timing of velocity sampling for the programming of saccades and pursuit in a step-ramp paradigm by briefly extinguishing the target at various times. Methods: Responses to horizontal “step-ramp” motions of a laser spot on a blank field were recorded with a dual-Purkinje eye tracker at 120 Hz. Nine step-ramp conditions were produced by combining −2, 0 and +2 deg steps with −8, 0 and +8 deg/s ramps. For each step-ramp, five target blanking conditions were used: either no gap or a 50-ms gap beginning at 25, 50, 75 or 100 ms after the step. All 45 conditions were presented in shuffled order in a single block of trials, and blocks were repeated 25 times for each of three subjects. Results: A 50-ms gap within the first 100 ms tended to increase the latency of the initial saccade. A gap starting at 25 ms reliably delayed pursuit onset when ramp motion was toward the fovea, whereas later gaps had little or no effect. With no gap, saccades were reliably larger in the step-ramp than the step-only condition, indicating that saccade programming incorporated ramp velocity information. Addition of a gap had no effect on saccade accuracy for step-only trials but often caused a relative undershoot in step-ramp trials. Steady state pursuit gain was unaffected by a gap. Conclusion: Gaps in velocity information generally affect saccades occurring 100 ms after gap onset. In contrast, only the earliest gap had an influence on pursuit latency, consistent with the shorter latency of these responses. The effects of gaps on saccade latency and position error are largely influenced by individual differences in oculomotor strategy adopted to track the stimulus.