N-methyl-d-aspartate (NMDA) receptor antagonists have been shown to protect against focal cerebral ischaemia when administered either before or soon after the onset of ischaemia. However, the precise therapeutic window for protection using these drugs remains to be defined. We studied dextrorphan administration delayed for 2 or 4 h after transient middle cerebral focal ischaemia in a rabbit model. With a 2h delay, the mid (12.5 mg kg-1 h-1) and high doses (17.5 mg kg-1 h-1) provided significant cortical neuroprotection (50% and 58% reduction, respectively), and the low dose (7.5 mg kg-1 h-1) protected against ischaemic damage in the basal ganglia (52% reduction). Animals having steady-state serum dextrorphan concentrations greater than 2000 ng ml-1 showed 50% cortical neuroprotection for the 2-h-delay group. No significant neuroprotection was seen in the 4-h-delay group, and the 4 h delay animals with dextrorphan levels greater than 2000 ng ml-1 had more severe ischaemic oedema than the saline controls. These results suggest a narrow temporal therapeutic window for neuroprotection, where delivery of drug delayed by 2 h was efficacious but treatment at 4 h after ischaemia onset was not beneficial and possibly harmful. These findings may have important implications for the treatment of clinical stroke.