In a defined Swedish population of 270,800, all patients with current chronic leg ulcers (827) were identified and a random sample of 382 studied in detail. Ulcers of primarily venous cause comprised 54 per cent of the total, giving a point prevalence of 0.16 per cent (95 per cent confidence interval 0.15-0.18 per cent). Half of all patients experienced their first ulcer episode before the age of 65 years, a greater proportion (61 per cent, P < 0.0001) in the subgroup with venous ulcers. The median duration of ulcer diathesis was significantly longer in patients with venous than in those with non-venous lesions (13.4 versus 2.5 years, P < 0.001). About half the patients with venous and non-venous ulcers had had their current lesion for longer than 1 year. Venous ulcers were more often recurrent than those of non-venous type (72 versus 45 per cent of patients, P < 0.0001). Patients with venous ulcers had a significantly higher body mass index (P < 0.001). The number of dressing changes performed per week was 1100 per 100,000 population. The predictive value of 'classical' clinical indicators of venous ulcer did not exceed 0.76. To increase the accuracy of diagnosis of venous ulcer, clinical examination should be combined with non-invasive 'objective' haemodynamic assessment of the venous circulation.