We have shown previously that pupil diameter increases and the amplitude of the pupillary light reflex is reduced when subjects are under threat of an aversive event (electric shock), and that light reflex amplitude correlates negatively with subjective anxiety. We have suggested that the "fear-inhibited light reflex" paradigm could be used as a laboratory model of human anxiety. In the present study, we examined whether two doses (5 mg and 10 mg) of the anxiolytic drug diazepam would antagonize the effects of threat on the pupillary light reflex. Twelve healthy male volunteers participated in three weekly sessions, each associated with one of three treatments (diazepam 5 mg or 10 mg or placebo) in a double-blind, balanced, cross-over design. The light reflex was recorded during either the anticipation of a shock ("threat" blocks) or periods in which no shocks were anticipated ("safe" blocks). At the end of each "threat" or "safe" block, subjects rated their anxiety using visual analogue scales. Two-factor ANOVA (treatment x condition) showed that diazepam treatment antagonized the effect of threat on light reflex amplitude in a dose-dependent manner but it did not affect the threat-induced increase in pupil diameter. Diazepam had no effect on the pupillary light reflex in the "safe" condition. Diazepam also reduced subjective anxiety and alertness in the threat condition. These results show the sensitivity of the threat-induced reduction of light reflex amplitude to anxiolytic drugs, and provide further evidence for the utility of the fear-inhibited light reflex paradigm as a laboratory model of human anxiety.