Studies in several laboratories have confirmed the anxiolytic potential of a wide range of 5-HT 1 A receptor antagonists in rats and mice, with recent evidence pointing to a postsynaptic site of action in the ventral hippocampus. It would, therefore, be predicted that blockade of 5-HT 1 A somatodendritic autoreceptors in the midbrain raphe nuclei should produce anxiogenic-like effects. To test this hypothesis, we investigated the effects of WAY-100635 microinfusions (0, 1.0 or 3.0 μg in 0.1 μl) into the dorsal (DRN) or median (MRN) raphe nuclei on behaviours displayed by male Swiss-Webster mice in the elevated plus-maze. As this test is sensitive to prior experience, the effects of intra-raphe infusions were examined both in maze-naive and maze-experienced subjects. Sessions were videotaped and subsequently scored for conventional indices of anxiety (open arm avoidance) and locomotor activity (closed arm entries), as well as a range of ethological measures (e.g. risk assessment). In maze-naive mice, intra-MRN (but not intra-DRN) infusions of WAY-100635 (3.0 μg) increased open arm exploration and reduced risk assessment. Importantly, these effects could not be attributed to a general reduction in locomotor activity. A similar, though somewhat weaker, pattern of behavioural change was observed in maze-experienced animals. This unexpected anxiolytic effect of 5-HT 1 A autoreceptor blockade in the MRN cannot be accounted for by a disinhibition of 5-HT release in forebrain targets (e.g. hippocampus and amygdala), where stimulation of postsynaptic 5-HT 1 A receptors enhances anxiety-like responses. However, as the MRN also projects to the periaqueductal gray matter (PAG), an area known to be sensitive to the anti-aversive effects of 5-HT, it is argued that present results may reflect increased 5-HT release at this crucial midbrain locus within the neural circuitry of defense.