Anxiolytic benzodiazepines have been shown to impair place learning in the Morris water maze. However, a clear-cut demonstration of a direct and specific effect on mnemonic processes has not yet been offered. In the present study, the effects of diazepam on place navigation in the Morris water maze were studied in rats. Three conditions were examined: learning, reversal learning and learning after familiarisation of animals with the maze. In view of the anxiolytic and sedative properties of diazepam, appropriate doses of the drug, i.e. those that produced an anxiolytic effect but no major motor impairment, were initially selected in the water-lick conflict and rotarod tests, respectively. Doses of 2.5 and 5 mg/kg PO increased punished drinking in the water-lick conflict test without significantly decreasing rotarod performance. These doses were then used to assess the effects of diazepam on spatial behaviour. Diazepam, at both doses, impaired place learning in behaviourally naive rats. Such an effect appeared to be transient: diazepam-treated rats eventually reached control performance. Moreover, analysis of the probe trial at the end of training revealed adoption of a spatial strategy to locate the submerged platform. Neither reversal learning nor learning after familiarisation was affected. These results do not replicate previous findings in the Morris water maze and provide some evidence that the diazepam-induced place learning deficit may be primarily anxiolytic in nature.