Saslow (1967) and Fischer and Ramsperger (1984) found that saccadic reaction time (SRT) depends on the interval between the fixation point offset and the target onset. Using a continuously visible fixation point, we asked whether a similar function would be obtained if subjects attended to a peripherally viewed point extinguished at variable intervals before or after the target onset. The interval was varied between -500 ms (i.e., attention stimulus offset after saccade target onset = overlap trials) and 500 ms (i.e., attention stimulus offset before saccade target onset = gap trials). The results show a constant mean SRT of about 240 ms for overlap trials, and a U-shaped function with a minimum of 140 ms, at a gap duration of 200 ms, for gap trials. These findings suggest that saccadic latencies do not depend on the cessation of fixation per se, but rather on the disengagement of attention from any location in the visual field. The time required for subjects to disengage their attention is approximately 100 ms. This disengaged state of attention--during which short latency (express) saccades can be made--can be sustained only for a gap duration of 300 ms. At longer gap durations mean SRTs increase again.