Little is known about the epidemiology of cardiac disease in diabetic end-stage renal disease. We therefore prospectively followed a cohort of 433 patients who survived 6 months after the inception of dialysis therapy for an average of 41 months. Clinical and echocardiographic data were collected yearly. At baseline, diabetic patients (n = 116) had more echocardiographic concentric left ventricular hypertrophy (50 vs 38%, p = 0.04), clinically diagnosed ischaemic heart disease (32 vs 18%, p = 0.003) and cardiac failure (48 vs 24%, p < 0.00001) than non-diabetic patients (n = 317). After adjusting for age and sex, diabetic patients had similar rates of progression of echocardiographic disorders, and de novo cardiac failure, but higher rates of de novo clinically diagnosed ischaemic heart disease (RR 3.2, p = 0.0002), overall mortality (RR 2.3, p < 0.0001) and cardiovascular mortality (RR 2.6, p < 0.0001) than non-diabetic patients. Mortality was higher in diabetic patients following admission for clinically diagnosed ischaemic heart disease (RR 1.7, p = 0.05) and cardiac failure (RR 2.2, p = 0.0003). Among diabetic patients older age, left ventricular hypertrophy, smoking, clinically diagnosed ischaemic heart disease, cardiac failure and hypoalbuminaemia were independently associated with mortality. The excessive cardiac morbidity and mortality of diabetic patients seem to be mediated via ischaemic disease, rather than progression of cardiomyopathy while on dialysis therapy. Potentially remediable risk factors include smoking, left ventricular hypertrophy, and hypoalbuminaemia.