To address the issue of whether single-photon emission computed tomography (SPECT) may add useful topographic information to neurologic examination in clinical practice, we compared technetium-99m hexamethyl propyleneamine oxime (HMPAO) SPECT and brain CT in 100 consecutive stroke patients involving the following clinical vascular territories: cortical middle cerebral artery (CMCA), 52; deep, 39; anterior cerebral artery (ACA), 3; posterior cerebral artery (PCA), 7. SPECT and CT sensitivity were calculated as the percentage of patients with respectively focal hypoperfusion or hypodensity related to the index event. The overall sensitivity of SPECT was not significantly different from CT (69 vs. 73%). The sensitivity of SPECT was not significantly dependent on the time of investigation (72% within 24 h, 50% on day 2, and 71 % after day 3). Compared with CT, the sensitivity of SPECT was statistically significantly higher in CMCA infarcts (89 vs. 71%; p = 0.027) and lower in deep infarcts (40 vs. 69%; p = 0.024). The degree of hypoperfusion was significantly higher in cortical lesions than in deep lesions (27 vs. 9%; p < 0.001). Crossed cerebellar diaschisis seemed less frequent in patients with clinical lesion in the ACA and PCA territory (respectively 30 and 40%) and was not significantly different when the ischemic lesion involved the CMCA (56%) or deep territories (50%). No significant difference was found when SPECT took place within or after 48 h. Thus, the selective SPECT sensitivity according to the site of lesion and the degree of hypoperfusion may help to determine stroke subtypes in the acute phase. In contrast, in the subacute phase, SPECT is as sensitive as CT in mapping the ischemic lesion and should not be recommended for this purpose.