A number of simple clinical and laboratory variables were analysed in a group of patients with chronic heart failure to evaluate their prognostic significance. Five hundred and fifty-two patients were followed for a maximum of 13 years with a total exposure time to death or censored survival of 1148 years. Of the clinical variables, diuretic dose and NYHA class were related to mortality (P < 0.01), and ischaemic heart disease was associated with a worse prognosis than other aetiologies (P < 0.05). Of the laboratory variables, abnormalities of liver function tests including bilirubin (P < 0.01), aspartate transaminase (P < 0.005), gamma glutamyl transpeptidase (P < 0.005) and alkaline phosphatase (P < 0.01) were all related to mortality as was plasma urate (P < 0.01). Multivariate survival analysis of all variables showed aspartate transaminase (chi 2 17.36, P < 0.001) accounted for the greatest variance followed by serum bilirubin (chi 2 14.35, P < 0.005). Thus, abnormalities in liver function tests have prognostic importance in chronic heart failure.