Diffusion-weighted magnetic resonance imaging (DWI) is a sensitive diagnostic tool for detecting recent ischemic lesions in patients with transient ischemic attacks (TIAs), but the interpretation of the presence or absence of DWI abnormalities in TIA patients still remains controversial. To elucidate the pathophysiology underlying those lesions, we analyzed DWI abnormalities in patients with recent TIAs. Based on 45 consecutive patients with TIAs who underwent DWI within 10 days of onset, demographic data and clinical manifestations were analyzed in relation to the DWI abnormalities. According to the method utilized in the Oxfordshire Community Stroke Study, clinical manifestations were classified into classical lacunar syndrome and non-lacunar symptoms. Based on the vascular distributions of ischemic lesions, the DWI abnormalities were classified into small-vessel and large-vessel lesions. DWI abnormalities were detected in 14 (31%) of 45 TIA patients. Seven (50%) of 14 DWI-positive patients had occlusive vascular lesions on intracranial magnetic resonance angiography, while only 5 (16%) of 31 DWI-negative patients had occlusive lesions (P < .05). No other demographic or clinical features, including risk factor and presence of cardiac disease, differed significantly between the DWI-positive and DWI-negative patient groups. Four (46%) of 9 DWI-positive patients who had a classical lacunar syndrome also showed small-vessel lesions on DWI, whereas all 5 patients who had non-lacunar symptoms showed large-vessel lesions. We concluded that although DWI abnormalities were detected in only one third of our TIA patients, DWI abnormalities were closely related to intracranial vascular occlusive lesions. The combination of DWI and MRA was useful for detecting large-artery lesions in patients displaying a classical lacunar syndrome.