1. Our goal was to use behavioral experiments to delimit where in the simian oculomotor system the gain of horizontal saccadic eye movements might be controlled. Our strategy was to change the gain of saccades to visual target steps (called targeting saccades) and to examine whether these changes transferred to other types of saccades. We reduced the gain of targeting saccades by jumping the target backward as a saccade was made so that the saccade appeared to overshoot. After 1,000-1,500 saccades to such backstepping targets, the average overshoot, and therefore the saccadic gain, had decreased substantially. 2. After the gain of targeting saccades had been reduced by 15-22%, several kinds of saccades were tested. Most were elicited by various visual targets. Some were made to jumping targets, which were timed to elicit saccades with longer (delayed saccades) or shorter (express saccades) latencies than normal or to targets that disappeared after a brief exposure (memory-guided saccades). Others were elicited to stationary targets (self-paced saccades) or in pursuit of a smoothly moving target (catchup saccades). Finally, we tested the saccadic fast phases of vestibular and optokinetic nystagmus. 3. Gain reduction of targeting saccades transferred at least partially to all the other types of saccades made to target jumps. The percentage gain transfer was calculated as (gain reduction of test saccades)/(gain reduction of adapted targeting saccades). The average percent transfer to delayed, memory-guided, and express saccades was 96, 88, and 91%, respectively. 4. Monkeys also showed substantial gain transfer to self-paced saccades, which scanned stationary targets. The average percentage gain transfer was 69% in the four animals tested. When two humans performed the same task, there was no transfer at all. These data suggest that saccadic gain adjustment involves different processes in monkeys and humans. 5. The transfer of gain to the catchup saccades of smooth pursuit varied from 41 to 100% across the four monkeys tested. Nevertheless, the average percentage gain transfer for all the animals was 75%. 6. As judged by the amplitude distribution of fast phases before and after adaptation, there was little, if any, saccadic gain transfer to the fast phases of vestibular or optokinetic nystagmus. In 12 of 13 experiments, there was no significant decrease in fast phase amplitude after a gain reduction of targeting saccades (P > 0.1). 7. This study shows that the average percentage gain transfer from targeting to delayed, express, memory-guided, self-paced, and catchup saccades was never < 69%. Although there was substantial transfer to saccades elicited by jumping, stationary, remembered, or slowly moving visual targets, there was relatively little to the saccadelike fast phases of nystagmus. The transfer of saccadic gain to the very short-latency express saccades suggests that adaptation modifies a subcortical locus. Moreover, the major locus must lie only in the premotor pathway for visual saccades, because saccadic gain adaptation is only poorly transferred to the fast phases of vestibular and optokinetic nystagmus.