Two rhesus monkeys were trained to track a small moving target in the presence of a moving distractor. The target and distractor were distinguished by their color. Smooth pursuit eye movements were quantified in terms of the latency of the eye movement and the open-loop eye acceleration profile. Smooth pursuit latencies for single targets were on the order of 100 msec. When the target was paired with a distractor moving in the same direction as the target, pursuit latencies decreased to roughly 85 msec. When the target was paired with a distractor moving in the opposite direction, pursuit latencies increased to roughly 150 msec. The motion of the distractor had no significant effect on the eye acceleration profile. Experiments were performed to dissociate visual search for the target from pursuit initiation by providing a spatial cue rather than the color cue. These experiments showed that visual search necessarily preceded pursuit initiation only when the distractor moved in the opposite direction relative to the target. In this case, visual search contributed about 25 msec to the overall latency of pursuit. Control experiments showed that the monkey need not attend to the distractor in order for it to influence the latency of pursuit. A network model was developed in which units that represent the motions of the target and distractor compete against one another. Attention serves to bias the outcome of this competition toward the direction of the selected target. The performance of this network exhibits a striking parallel to the effect of the distractor on smooth pursuit latency.