This report describes an experiment in which was measured the onset latency of the saccade to the peripheral target under three conditions of visual cues for the target location. In the central cue condition (the C-condition), the cue was an arrow presented in the central visual field. In the peripheral cue condition (the P-condition), the cue was the form in which the target was presented in the peripheral visual field. In the one-sided cue condition (the O-condition), the cue was the target itself. The saccade latency was measured by means of electro-oculography. The results were; (1) the saccade latencies in the three conditions were significantly different from each other, (2) the latency was longest in the C-condition and shortest in the O-condition, (3) the difference in latency between the P and C-conditions was three times as large as that between O and P-conditions, (4) the latency variance was significantly larger in the C-condition than in the other conditions, (5) the difference in the variances was not significant between the O and P-conditions. The results were interpreted in terms of (a) cognitive processing of visual form and (b) the characteristics of the saccade system.