1. The dynamic contribution of otolith signals to three-dimensional angular vestibuloocular reflex (VOR) was studied during off-vertical axis rotations in rhesus monkeys. In an attempt to separate response components to head velocity from those to head position relative to gravity during low-frequency sinusoidal oscillations, large oscillation amplitudes were chosen such that peak-to-peak head displacements exceeded 360 degrees. Because the waveforms of head position and velocity differed in shape and frequency content, the particular head position and angular velocity sensitivity of otolith-ocular responses could be independently assessed. 2. During both constant velocity rotation and low-frequency sinusoidal oscillations, the otolith system generated two different types of oculomotor responses: 1) modulation of three-dimensional eye position and/or eye velocity as a function of head position relative to gravity, as presented in the preceding paper, and 2) slow-phase eye velocity as a function of head angular velocity. These two types of otolith-ocular responses have been analyzed separately. In this paper we focus on the angular velocity responses of the otolith system. 3. During constant velocity off-vertical axis rotations, a steady-state nystagmus was elicited that was maintained throughout rotation. During low-frequency sinusoidal off-vertical axis oscillations, dynamic otolith stimulation resulted primarily in a reduction of phase leads that characterize low-frequency VOR during earth-vertical axis rotations. Both of these effects are the result of an internally generated head angular velocity signal of otolithic origin that is coupled through a low-pass filter to the VOR. No change in either VOR gain or phase was observed at stimulus frequencies larger than 0.1 Hz. 4. The dynamic otolith contribution to low-frequency angular VOR exhibited three-dimensional response characteristics with some quantitative differences in the different response components. For horizontal VOR, the amplitude of the steady-state slow-phase velocity during constant velocity rotation and the reduction of phase leads during sinusoidal oscillation were relatively independent of tilt angle (for angles larger than approximately 10 degrees). For vertical and torsional VOR, the amplitude of steady-state slow-phase eye velocity during constant velocity rotation increased, and the phase leads during sinusoidal oscillation decreased with increasing tilt angle. The largest steady-state response amplitudes and smallest phase leads were observed during vertical/torsional VOR about an earth-horizontal axis. 5. The dynamic range of otolith-borne head angular velocity information in the VOR was limited to velocities up to approximately 110 degrees/s. Higher head velocities resulted in saturation and a decrease in the amplitude of the steady-state response components during constant velocity rotation and in increased phase leads during sinusoidal oscillations. 6. The response characteristics of otolith-borne angular VORs were also studied in animals after selective semicircular canal inactivation. Otolith angular VORs exhibited clear low-pass filtered properties with a corner frequency of approximately 0.05-0.1 Hz. Vectorial summation of canal VOR alone (elicited during earth-vertical axis rotations) and otolith VOR alone (elicited during off-vertical axis oscillations after semicircular canal inactivation) could not predict VOR gain and phase during off-vertical axis rotations in intact animals. This suggests a more complex interaction of semicircular canal and otolith signals. 7. The results of this study show that the primate low-frequency enhancement of VOR dynamics during off-vertical axis rotation is independent of a simultaneous activation of the vertical and torsional "tilt" otolith-ocular reflexes that have been characterized in the preceding paper. (ABSTRACT TRUNCATED)