BACKGROUND Stringent transcranial Doppler (TCD) criteria for diagnosing occlusion are needed for more reliable TCD performance at bedside in the acute stroke setting. SUBJECTS AND METHODS At three academic stroke centers, we performed TCD examination for patients with symptoms of cerebral ischemia who underwent digital subtraction angiography (DSA). We used a standard insonation protocol with power M-mode Doppler (PMD) TCD (TCD 100 M, Spencer Technologies Inc., Seattle, WA). We collected mean flow velocity (MFV), pulsatility indices (PI), and power M-mode resistance signature (absent, high, or low) in symptomatic middle (MCA), anterior (ACA), posterior (PCA), and in affected (a), ipsilateral (i), and contralateral (c-lat) cerebral arteries. Ratios of aMCA/c-lat MCA, aMCA/iACA, and aMCA/iPCA MFV were subsequently calculated. PMD-TCD flow findings were evaluated with a receiver-operating characteristic (ROC) analysis for angiographically proven MCA occlusion. RESULTS We studied 120 patients with acute cerebral ischemia with PMD-TCD examinations prior to or immediately after DSA. Lower aMCA velocities pointed to higher probability of occlusion (P= .055). The aMCA/iPCA MFV ratio was superior to the aMCA/iACA ratio and strongly predictive of occlusion at a threshold ratio of 0.5 (RR 2.31 CI(95) 2.13-2.51). High resistance or absent M-mode flow signatures in the proximal MCA were present in 87% of M1 and M2 MCA occlusions (probability 87%). In the presence of a low-resistance PMD signature, obtaining the aMCA/iPCA MFV ratio <0.5 increases probability of occlusion to 87%. Normal MFV ratios and low-resistance M-mode signatures are highly predictive of a negative angiogram for MCA occlusion. CONCLUSION In acute cerebral ischemia, reliable criteria for proximal MCA occlusion have been developed based on combination of MFV ratios and M-mode flow resistance signatures. Validation of these criteria will require multicenter studies.