Rats were given 10 light-shock pairings on 2 successive days. At 24-48 hr following training, groups of rats received bilateral transection of the cerebellar peduncles, bilateral lesions of the red nucleus (which receives most of the cerebellar efferents), or bilateral lesions of the central nucleus of the amygdala. Control rats were sham operated. At 3-4 days after surgery, the rats were tested for potentiated startle (increased acoustic startle in the presence of the light previously paired with shock). Potentiated startle was blocked by lesions of the central nucleus of the amygdala. Transection of the cerebellar peduncles or lesions of the red nucleus did not block potentiated startle. A second experiment in which a visual prepulse test was used indicated that the blockade of potentiated startle observed in the animals with amygdala lesions could not be attributed to optic tract damage. A third experiment demonstrated that the absence of potentiation in the animals with amygdala lesions was not simply due to a lowered startle level ceiling, because these animals could show increased startle with increased stimulus intensity and with administration of strychnine, a drug that increases startle. Taken together, the results are consistent with the hypothesis that the amygdala is involved in fear conditioning, because potentiated startle is a measure of conditioned fear.