Several studies suggest that cyclooxygenase (COX)-2 plays a pivotal role in the progression of ischaemic brain damage. In the present study, we investigated the effects of selective inhibition of COX-2 with nimesulide (12 mg/kg) and selective inhibition of COX-1 with valeryl salicylate (VAS, 12–120 mg/kg) on prostaglandin E2 (PGE2) levels, myeloperoxidase (MPO) activity, Evans blue (EB) extravasation and infarct volume in a standardized model of transient focal cerebral ischaemia in the rat. Post-ischaemic treatment with nimesulide markedly reduced the increase in PGE2 levels in the ischaemic cerebral cortex 24 h after stroke and diminished infarct size by 48% with respect to vehicle-treated animals after 3 days of reperfusion. Furthermore, nimesulide significantly attenuated the blood–brain barrier (BBB) damage and leukocyte infiltration (as measured by EB leakage and MPO activity, respectively) seen at 48 h after the initial ischaemic episode. These studies provide the first experimental evidence that COX-2 inhibition with nimesulide is able to limit BBB disruption and leukocyte infiltration following transient focal cerebral ischaemia. Neuroprotection afforded by nimesulide is observed even when the treatment is delayed until 6 h after the onset of ischaemia, confirming a wide therapeutic window of COX-2 inhibitors in experimental stroke. On the contrary, selective inhibition of COX-1 with VAS had no significant effect on the evaluated parameters. These data suggest that COX-2 activity, but not COX-1 activity, contributes to the progression of focal ischaemic brain injury, and that the beneficial effects observed with non-selective COX inhibitors are probably associated to COX-2 rather than to COX-1 inhibition.