The present study determined cerebral blood flow (CBF) in the rat using two different magnetic resonance imaging (MRI) arterial spin-tagging (AST) methods and 14C-iodoantipyrine (IAP)-quantitative autoradiography (QAR), a standard but terminal technique used for imaging and quantitating CBF, and compared the resulting data sets to assess the precision and accuracy of the different techniques. Two hours after cerebral ischemia was produced in eight rats via permanent occlusion of one middle cerebral artery (MCA) with an intraluminal suture, MRI-CBF was measured over a 2.0-mm coronal slice using single-coil AST, and tissue magnetization was assessed by either a spin-echo (SE) or a variable tip-angle gradient-echo (VTA-GE) readout. Subsequently ( approximately 2.5 hours after MCA occlusion), CBF was assayed by QAR with the blood flow indicator 14C-IAP, which produced coronal images of local flow rates every 0.4 mm along the rostral-caudal axis. The IAP-QAR images that spanned the 2-mm MRI slice were selected, and regional flow rates (i.e., local CBF [lCBF]) were measured and averaged across this set of images by both the traditional approach, which involved reader interaction and avoidance of sectioning artifacts, and a whole film-scanning technique, which approximated total radioactivity in the entire MRI slice with minimal user bias. After alignment and coregistration, the concordance of the CBF rates generated by the two QAR approaches and the two AST methods was examined for nine regions of interest in each hemisphere. The QAR-lCBF rates were higher with the traditional method of assaying tissue radioactivity than with the MRI-analog approach; although the two sets of rates were highly correlated, the scatter was broad. The flow rates obtained with the whole film-scanning technique were chosen for subsequent comparisons to MRI-CBF results because of the similarity in tissue "sampling" among these three methods. As predicted by previous modeling, "true" flow rates, assumed to be given by QAR-lCBF, tended to be slightly lower than those measured by SE and were appreciably lower than those assessed by VTA-GE. When both the ischemic and contralateral hemispheres were considered together, SE-CBF and VTA-GE-CBF were both highly correlated with QAR-lCBF ( P< 0.001). If evaluated by flow range, however, SE-CBF estimates were more accurate in high-flow (contralateral) areas (CBF > 80 mL. 100 g(-1). min(-1) ), whereas VTA-GE-CBF values were more accurate in low-flow (ipsilateral) areas (CBF < or= 60 mL. 100 g(-1). min(-1) ). Accordingly, the concurrent usage of both AST-MRI methods or the VTA-GE technique alone would be preferred for human studies of stroke.