We studied the interaural phase sensitivity of 85 units in the inferior colliculus (IC) of the unanesthetized rabbit. We assessed this sensitivity at several frequencies within each unit's responsive range. The interaural phase disparity was varied by delivering tones that differed by 1 Hz to the two ears, resulting in a 1-Hz binaural beat. We analyzed each unit's response to different frequencies by calculating four measures: characteristic delay (CD), characteristic phase (CP), composite peak delay, and mean peak delay. We estimated the CD and CP from the slope and phase intercept, respectively, of the regression line fitted to a plot of the mean interaural phase against stimulating frequency. The composite peak delay was estimated from the peak of a composite delay curve. This was generated by replotting the response to changes in interaural phase, as a function of the equivalent interaural delay and averaging the resultant interaural delay curves. The composite delay curve reflects the unit's average response to interaural delays across frequencies. Last, we calculated a mean peak delay, derived by converting the mean interaural phase of the response at each frequency to an equivalent delay and then averaging these delays. Interaural phase sensitivity was observed to frequencies as high as 2,150 Hz. However, the majority of units showed such sensitivity below 1,500 Hz. For most units, the interaural delay curves measured at several frequencies coincided near the peak discharge. This result is consistent with a neural model, where excitatory inputs from each ear converge upon a binaural cell, evoking maximum discharge only when the two inputs arrive simultaneously. As a first approximation, our data fit this model, indicating that IC neurons can act like coincidence detectors or cross-correlators. The distributions of CD, composite peak delay, and mean peak delay showed that most units preferred ipsilateral stimulus delays, which in the natural situation corresponds to sounds emanating from the contralateral field. Moreover, most units preferred delays that were within the estimated physiological range of the rabbit. These results support the viewpoint that neurons in the IC participate in sound localization. The distributions of CP and CD differ substantially from those found in the IC of the anesthetized cat. These differences may reflect species differences, the effects of anesthesia, or a difference in the population of units sampled. For each unit, we assessed the linearity of the plot of mean interaural phase against frequency of stimulation using a chi 2 method. For most units the plots were significantly nonlinear.(ABSTRACT TRUNCATED AT 400 WORDS)