Objective: To establish the validity of visual interpretation of immediately processed perfusion computed tomography (CT) maps in acute stroke for prediction of final infarction. Methods: Perfusion CT studies acquired prospectively were reprocessed within six hours of stroke onset using standard CT console software. Four contiguous 5 mm thick images were obtained and maps of time to peak (TTP) and cerebral blood volume (CBV) generated. Volumes of lesions identified only by visual inspection were measured from manually drawn regions of interest. Volumes of tissue with prolonged TTP or reduced CBV were compared with independently calculated volume of infarction on non-contrast CT (NCCT) at 24-48 hours, and with clinical severity using the NIHSS score. Arterial patency at 24-48 h was included in analyses. Results: Studies were analysed from 17 patients 150 minutes (median) after stroke onset. Volume of tissue with prolonged TTP correlated with initial NIHSS (r = 0.62, p = 0.009), and with NCCT final infarct volume when arterial occlusion persisted (r = 0.953, p = 0.012). Volume of tissue with reduced CBV correlated with final infarct volume if recanalisation occurred (r = 0.835, p = 0.001). Recanalisation was associated with lower 24 h NIHSS score (6 (IQR, 5 to 9.5) v 19 (18 to 26), p = 0.027), and in 10 patients given rtPA for MCA M1 occlusion, with lower infarct volume (73 v 431 ml, p = 0.002). Conclusions: Visual evaluation of TTP and CBV maps generated by standard perfusion CT software correlated with 24-48 hour CT infarct volumes. Comparison of TTP and CBV maps yields information on tissue viability. Perfusion CT represents a practical technique to aid acute clinical decision making. Recanalisation was a crucial determinant of clinical and radiological outcome.