1. The introduction of a period of darkness between the disappearance of an initial fixation target and the appearance of a peripheral saccade target produces a general reduction in saccadic reaction time (SRT)-known as the gap effect- and often very short latency express saccades. To account for these phenomena, premotor processes may be facilitated by release of visual fixation and advanced preparation of saccadic programs. The experiments described in this paper were designed to test the relevance of the ocular fixation disengagement and oculomotor preparation hypotheses by identifying the influence of different factors on SRTs and the occurrence of express saccades in the monkey. 2. The SRTs of two monkeys were measured in two behavioral paradigms. A peripheral saccade target appeared at the time of disappearance of a central fixation target in the no-gap task, whereas a 200-ms period of no stimuli was interposed between the fixation target disappearance and the saccade target appearance in the gap task. The distribution of SRTs in these tasks was generally bimodal; the first and second mode was composed of express and regular saccades, respectively. We measured the mean SRT, mean regular saccade latency, mean express saccade latency, and percentage of express saccades in both tasks. We also estimated the gap effect, i.e., the difference between the SRTs in no-gap trial and the SRTs in gap trials. 3. Once the animals were trained to make saccades to a single target location and produce express saccades, SRTs in both no-gap and gap trials displayed a broad tuning with respect to the spatial location of the trained target when the target location was varied randomly in a block of trials. Express saccades were made only to a restricted region of the visual field surrounding the trained target location. A gap effect was present for nearly all target locations tested, irrespective of express saccade occurrence. Finally, the probability of generating an express saccade at the trained target location decreased with the introduction of uncertainty about target location. 4. The occurrence of express saccades increased with the duration of the visual and nonvisual (gap) fixation that the animal was required to maintain before the onset of a saccade target. The gap duration was effective in reducing the mean SRT for gaps < or = 300 ms, and it was more influential than comparable variation in the visual fixation duration. 5. The occurrence of express saccades made to targets of identical eccentricity increased when the initial eye fixation position was shifted eccentric in a direction opposite to the saccade direction. Concomitantly, mean SRT decreased by approximately 2 ms for each 1-deg change in initial eye fixation position. 6. The occurrence of express saccades depended upon contextual factors, i.e., on both the behavioral task (no-gap or gap) and the latency of the saccade that the monkey executed to the same target in the preceding trial. The highest percentage of express saccades was observed after an express saccade in a no-gap trial, whereas the lowest percentage was obtained after a regular saccade in a gap trial. 7. These findings indicate that training-dependent express saccades are restricted to a specific spatial location dictated by the training target, and their incidence is facilitated by high predictability of target presentation, long-duration foreperiod, absence of visual fixation, eccentric initial eye position opposite to the saccade direction, and express saccade occurrence in the previous trial. The release of fixation afforded by the gap accounts for the general gap effect, but has only a modulatory influence on express saccade generation. We conclude that advanced motor preparation of saccadic programs generally reduces SRT and is primarily responsible for the occurrence of express saccades, which therefore may be caused mainly by neuronal changes restricted to a specific locus-coding for the trained movemen