1 The neuroprotective properties of fasudil (HA1077), a novel protein kinase inhibitor, were evaluated in two animal models of cerebral ischaemia: transient bilateral carotid artery occlusion in Mongolian gerbils and cerebral microembolization in rats. 2 The cytoprotective effect of fasudil on delayed neuronal death in gerbils was compared with the effects of nimodipine, a calcium channel antagonist and ozagrel, a thromboxane A2 synthetase inhibitor.The average of the neuronal cell density in the ischaemic control group was 17.8+2.1 cells mm-', whereas fasudil (30 mg kg-') significantly diminished the loss of CAl neurones with the average of the neuronal cell density of 101.0+22.0 cells mm-'; nimodipine (10 mg kg-') and ozagrel (30 mg kg-1) did not significantly protect against the ischaemia-induced neuronal loss. 3 In the rat model, the effects of fasudil on the histological and neurological consequences of cerebral microembolization produced via the injection of microspheres were examined. Twenty-four hours after the injection of microspheres into the internal carotid artery, all animals in the control group showed typical symptoms of stroke. Neurological function was significantly improved in the fasudil-treated animals. In the controls, the infarcted area in a cortical slice selected to include the hippocampal area was 0.25 + 0.01 cm2 (mean+s.e.mean) (43.9+2.4% of cortical section of the half hemisphere); the difference was significant compared to the mean area of 32.7+2.8 and 21.5+4.8% observed in rats treated with fasudil (3, 10 mg kg-'), respectively. Fasudil (10 mg kg-') significantly suppressed the increased water content in ischaemic brain tissues (saline-treated rats, 82.4+0.2% vs fasudil-treated rats, 81.0 + 0.4%). 4 These results suggest that: (i) various protein kinases are involved in the pathogenesis of ischaemic injury; and (ii) the inhibition of protein kinases may be efficacious in preventing neuronal death, thus improving neurological function in the brain damage associated with ischaemic stroke.