Acute administration of diazepam (2 mg/kg i.p.) to rats decreased the number of head-dips, locomotor activity and the number of rears made in the holeboard apparatus, indicating sedative effects. After daily treatment for 7 days with diazepam (2 mg/kg) tolerance developed to all these behavioural effects, despite serum concentrations of diazepam andN-desmethyldiazepam significantly higher than those following acute treatment. After 7 drug-free recovery days the rats were quite unresponsive to a probe dose of diazepam (2 mg/kg) and although there was a gradual recovery of responsiveness to diazepam, the reduction in rears still did not reach the level of the acute group even after 21 drug-free days. There was evidence for pharmacokinetic changes when probe doses of diazepam were given after 7, 14 or 21 recovery days. Lower levels of diazepam and higher levels ofN-desmethyldiazepam than following an acute dose to the drug-naive group were detected, indicating that the chronic treatment had resulted in a persistently enhanced rate ofN-demethylation. It is argued that these changes do not fully account for the reduced responsiveness to the probe doses, and nor can they account for the gradual return of response over the 3-week recovery period. There were no detectable serum concentrations of either compound 24 h after the end of the chronic treatment. However, no rebound increases in behavioural responses were detected at any time-point in withdrawal. Thus, the mechanism underlying this behavioural tolerance was not oppositional in nature. It is suggested that a situation-independent learned behavioural strategy is the most likely mechanism for the observed tolerance.