Responses of the inferior colliculus (IC) neurons to acoustical stimulation were recorded in anesthetized guinea pigs. Three kinds of anesthesia were used: (1) urethane (8 ml/kg b.w. of 20 per cent solution i.p.); (2) ketamine-xylazine combination (1 ml/kg b.w. of mixture 2:1); and (3) pentobarbital (25 mg/kg i.p.) combined with intramuscular injection of fentanyl (0.5 ml) and droperidol (1 ml). The frequency tuning of neurons evaluated on the basis of Q10 values and the composite neural audiogram represented by points of lowest thresholds of individual IC neurons were similar for guinea pigs treated with any of the anesthetics. The number of spontaneously active IC neurons was significantly larger with ketamine than with urethane or pentobarbital. The response latencies to tone bursts at characteristic frequency (CF) were shortest in animals anesthetized with pentobarbital. Whereas with ketamine and urethane many neurons were recorded in which response latencies were longer than 40 ms, in pentobarbital-anesthetized animals the latencies usually did not exceed 25 ms. The occurrence of neurons with an onset type of response was significantly larger with pentobarbital than with the other two anesthetics. In ketamine and urethane anesthesia, thresholds of units with sustained response were significantly lower than thresholds of units with onset response. Our results suggest that in experiments where the level of spontaneous activity, latency and type of responses were important parameters, the kind of anesthesia should be taken into account.