Previous findings suggest differences in the neuroanatomical substrates of short- (seconds) vs longer-duration (minutes) fear responses. We now report that phasic and sustained fear can also be differentiated pharmacologically, based on their response to several treatments that either are or are not clinically effective anxiolytics. For these experiments, short- or long-duration clicker stimuli were paired with footshock. Acoustic startle amplitude was later measured in the absence of the clicker, or within seconds (phasic fear) or minutes (sustained fear) of its onset. Before testing, rats received a single injection of vehicle, the benzodiazepine chlordiazepoxide, the 5HT(1A) agonist and dopamine D2 antagonist buspirone, the selective serotonin reuptake inhibitor fluoxetine, or a 3-week treatment with either vehicle or fluoxetine. Chlordiazepoxide blocked sustained, but not phasic startle increases. Acute buspirone, which is not anxiolytic in human beings, did not affect sustained startle increases, but did disrupt phasic increases. Chronic fluoxetine blocked sustained startle increases and unreliably reduced phasic increases; acute fluoxetine affected neither. The results indicate that phasic and sustained fear responses can be pharmacologically dissociated, further validating this distinction, and suggest that sustained startle increases may be especially useful as anxiety models and anxiolytic screens.