Caloric nystagmus of 18 normal subjects and 10 monkeys were compared three dimensionally. Eye movements in darkness were recorded by an electronystagmography (ENG) in humans and magnetic search coils in eight rhesus monkeys. In some of the humans and also in two japanese monkeys, eye movements were observed under infra-red video monitoring. Caloric stimuli were applied by irrigating the external ear canal with 20°C water in the ear up position; then the head position was changed to the supine or to the prone position. The first phase of caloric nystagmus had three dimensional components except the torsional component in the supine position in humans; i.e. the horizontal component beat away from the side of irrigation in the supine position, and toward the side of irrigation in the prone position. The torsional component was counterclockwise with right ear irrigation and clockwise with left ear irrigation in both the supine and prone positions; the vertical component was up beating in the prone position and down beating in the supine in both the right and left ear irrigations. The torsional second phase of caloric nystagmus in monkeys was very active and lasted quite long in both the supine and the prone positions. In the side position, torsional nystagmus was markedly suppressed and strong vertical nystagmus appaered. In humans, almost no second phase of caloric nystagmus was recorded in the supine or prone positions, while weak but definite horizontal nystagmus was recorded in every subjects in the sitting position. These findings of the second phase of caloric nystagmus strongly suggested that the second phase of caloric nystagmus appeared in the earth horizontal plane, where no otolithic control exists.