The prognosis, and clinical findings related to prognosis, were examined in a consecutive series of 190 patients under 76 years of age (mean 64 years) with congestive heart failure (CHF). The aetiology of CHF was ischaemic heart disease in 66%, hypertension in 11% and cardiomyopathy in 23%. The 2-year mortality was 32%. Median left ventricular ejection fraction (LVEF) was 0.30, range 0.06 to 0.74. Eight per cent were in New York Heart Association (NYHA) class I, 46% in II, 44% in III and 2% in IV. Multivariate analysis, excluding exercise test variables, revealed seven variables with independent, significant prognostic information, (hazard ratios for death in brackets): ln (natural logarithm) (LVEF) (3.19), NYHA class III+IV (2.72), plasma urea > 7.6 mmol.l-1 (2.22), serum creatinine > 121 mumol.l-1 (2.05), serum sodium < or = 137 mmol.l-1 (2.03), pulmonary congestion on X-ray (1.86), and age > 65 years (1.86). Multivariate analysis, including exercise testing, showed the following variables to contain independent, significant prognostic information: increase in heart rate during maximal exercise < or = 35 min-1 (3.5), ln (LVEF) (3.7), serum creatinine > 121 mumol.l-1 (2.9), maximal exercise time < or = 4 min (2.3), serum sodium < or = 137 mmol.l-1 (2.5), ischaemic heart disease (2.0) and plasma urea > 7.6 mmol.l-1 (1.9). In conclusion, patients with CHF have a high risk of death despite intensive medical treatment. LVEF is a strong predictor of mortality. Both NYHA class and exercise variables have strong independent prognostic information as regards mortality in combination with LVEF, but are mutually exclusive.