1. The discharge of antidromically identified medial rectus and abducens motoneurons was recorded in restrained unanesthesized goldfish during spontaneous eye movements and in response to vestibular and optokinetic stimulation. 2. All medial rectus and abducens motoneurons exhibited a similar discharge pattern. A burst of spikes accompanied spontaneous saccades and fast phases during vestibular and optokinetic nystagmus in the ON-direction. Firing rate decreased for the same eye movements in the OFF-direction. All units showed a steady firing rate proportional to eye position beyond their recruitment threshold. 3. Motoneuronal position (ks) and velocity (rs) sensitivity for spontaneous eye movements were calculated from the slope of the rate-position and rate-velocity linear regression lines, respectively. The averaged ks and rs values of medial rectus motoneurons were higher than those of abducens motoneurons. The differences in motoneuronal sensitivity coupled with structural variations in the lateral versus the medial rectus muscle suggest that symmetric nasal and temporal eye movements are preserved by different motor unit composition. Although the abducens nucleus consists of distinct rostral and caudal subgroups, mean ks and rs values were not significantly different between the two populations. 4. Every abducens and medial rectus motoneuron fired an intense burst of spikes during its corresponding temporal or nasal activation phase of the "eye blink." This eye movement consisted of a sequential, rather than a synergic, contraction of both vertical and horizontal extraocular muscles. The eye blink could act neither as a protective reflex nor as a goal-directed eye movement because it could not be evoked in response to sensory stimuli. We propose a role for the blink in recentering eye position. 5. Motoneuronal firing rate after ON-directed saccades decreased exponentially before reaching the sustained discharge proportional to the new eye position. Time constants of the exponential decay ranged from 50 to 300 ms. Longer time constants after the saccade were associated with backward drifts of eye position and shorter time constants with onward drifts. These postsaccadic slide signals are suggested to encode the transition of eye position to the new steady level. 6. Motoneurons modulated sinusoidally in response to sinusoidal head rotation in the dark, but for a part of the cycle they went into cutoff, dependent on their eye position recruitment threshold. Eye position (kv) and velocity (rv) sensitivity during vestibular stimulation were measured at frequencies between 1/16 and 2 Hz. Motoneuronal time constants (tau v = rv/kv) decreased on the average by 25% with the frequency of vestibular stimulation.(ABSTRACT TRUNCATED AT 400 WORDS)