Three components of pre-saccadic evoked potentials have been identified in humans: a slow negative shift (SNS), a positive antecedent potential (AP) and a spike potential (SP). This study examined the influences of: instructions to the subject to make saccades; the presence of a visual target; and the direction of the saccades on the amplitude of the averaged SP, which was recorded from P3 and P4 (International 10/20 System) in 20 normal, right-handed subjects. Recordings were made for spontaneous saccades prior to receiving instructions in six subjects. Twenty subjects performed self-paced saccades in the presence of a 10 degrees visual target (two red LEDs) and while blindfolded in a dark room. The SP was either absent or grossly altered (broadened) for spontaneous saccades in an illuminated room; it was robust for self-paced saccades in light or darkness. Three-way analysis of variance revealed a highly significant cortical laterality (P3 vs P4) X saccade direction interaction (P less than 0.001), reflecting that for a given saccade direction, the SP was larger over the contralateral recording site for the self-paced light (SPL) paradigm (Newman-Keuls test). In the self-paced dark (SPD) paradigm, though, this was only true for saccades to the right. By contrast, scatter plots of the directional indices (D.I. = [(SP for contralateral) - (SP for ipsilateral)]/[(SP for contralateral) + (SP for ipsilateral saccades)] for left (P3) and right (P4) recording sites from individual subjects revealed a significant negative correlation for both SPL (r = 0.78) and SPD (r = 0.74) paradigms.(ABSTRACT TRUNCATED AT 250 WORDS)