Background and Purpose-Transcranial Doppler (TCD) is used for diagnosis of vasospasm in patients with subarachnoid hemorrhage due to a ruptured aneurysm. Our aim was to evaluate both the accuracy of TCD compared with angiography and its usefulness as a screening method in this setting. Methods-A search (MEDLINE, EMBASE, Cochrane Library, bibliographies, hand searching, any language, through January 31, 2001) was performed for studies comparing TCD with angiography. Data were critically appraised using a modified published 10-point score and were combined using a random-effects model. Results-Twenty-six reports compared TCD with angiography. Median validity score was 4.5 (range 1 to 8). Metaanalyses could be performed with data from 7 trials. For the middle cerebral artery (5 trials, 317 tests), sensitivity was 67% (95% CI 48% to 87%), specificity was 99% (98% to 100%), positive predictive value (PPV) was 97% (95% to 98%), and negative predictive value (NPV) was 78% (65% to 91%). For the anterior cerebral artery (3 trials, 171 tests), sensitivity was 42% (11% to 72%), specificity was 76% (53% to 100%), PPV was 56% (27% to 84%), and NPV was 69% (43% to 95%). Three of these 7 studies reported on the same patients, each on another artery, and for 4, data recycling could not be disproved. Other arteries were tested in only 1 trial each. Conclusions-For the middle cerebral artery, TCD is not likely to indicate a spasm when angiography does not show one (high specificity), and TCD may be used to identify patients with a spasm (high PPV). For all other situations and arteries, there is either lack of evidence of accuracy or of any usefulness of TCD. Most of these data are of low methodological quality, bias cannot not be ruled out, and data reporting is often uncritical. (Stroke. 2001;32:2292-2298.)