The incidence and vascular type of recurrent ischaemic stroke was studied in patients with supratentorial transient ischaemic attacks or non-disabling ischaemic strokes, who were treated with aspirin (30 or 283 mg). Patients were divided into groups with small vessel disease (SVD) (n = 1216) or large vessel disease (LVD) (n = 1221) on the grounds of their clinical features and CT at baseline. Patients with evidence of both SVD and LVD (n = 180) were excluded from further analyses. During follow up (mean 2.6 years) annual stroke rate was 3.6% in both groups. Of the 107 patients with SVD at baseline who had recurrent strokes, 83 proved to have an identifiable infarct: 30 (28%) again had a small vessel infarct, 39 (36%) had a large vessel ischaemic stroke and in 14 (13%) the recurrent ischaemic stroke was in the posterior fossa. Of the 110 patients with LVD at baseline and recurrent stroke, 91 had an identifiable infarct: 67 (61%) again had a large vessel ischaemic stroke, 16 (15%) had a small vessel ischaemic stroke, and eight (7%) had the recurrent ischaemic stroke in the posterior fossa. Thus patients with a transient ischaemic attack or non-disabling ischaemic stroke caused by LVD were more likely to have an ischaemic stroke of the same vessel type during follow up than patients with SVD (relative risk 2.2; 95% confidence interval 1.5-3.4). Possible explanations for this difference are: (1) patients with a small vessel ischaemic stroke at baseline had both SVD and LVD or were misdiagnosed; (2) recurrent small vessel ischaemic stroke may have occurred more often than reported, because they were silent or only minimally disabling; (3) recurring large vessel ischaemic strokes occurring in patients initially diagnosed as having SVD might have been related to potential cardiac sources of emboli that had not been previously recognized; (4) the antiplatelet drug aspirin (30 or 283 mg) prescribed in this patient group may have prevented thrombosis in small vessels better than in large vessels.