Intracerebroventricular infusion of corticotropin-releasing factor (CRF) (0.1–1.0 micrograms) produced a pronounced, dose-dependent enhancement of the acoustic startle reflex in rats. This excitatory effect began about 20–30 min after infusion, grew steadily over the 2 hr test period, and lasted at least 6 hr. Higher doses of CRF (10 micrograms) often produced marked facilitation and then inhibition of startle that oscillated repeatedly with a period of 10–20 min. CRF- enhanced startle did not result from an increase in sensitization produced by repetition of the startle stimulus or from a blockade of habituation. Peripheral injections of the autonomic ganglionic blockers hexamethonium (10 mg/kg) or chlorisondamine (3 mg/kg) slightly attenuated the magnitude of CRF-enhanced startle, suggesting a partial role of peripheral sympathetic activation. Intracerebroventricular infusion of the CRF antagonist alpha-helical CRF9–41 (alpha hCRF; 25 or 50 micrograms) blocked CRF-enhanced startle when infused 5 min prior to CRF, indicating a central site of action. CRF-enhanced startle also was reversed when alpha hCRF was given 90 min after infusion of CRF. This suggests that exogenously applied CRF remains in the brain for a very long time after administration or that CRF given exogenously initiates a process that results in a long-lasting activation of endogenous CRF. Because the startle reflex is elevated by both conditioned and unconditioned fear, these data lend further support to the idea that CRF infusion produces a behavioral state that resembles fear or anxiety. Because startle is mediated by a well-defined neural pathway, CRF-enhanced startle may provide a useful behavioral assay to analyze the neural systems upon which exogenous CRF acts to produce its behavioral effects.