OBJECTIVES--This study examines the acute effects of two differing beta adrenergic blocking agents (metoprolol and a third generation vasodilating beta blocker) on plasma concentrations of atrial natriuretic factor (ANF), brain (ventricular) natriuretic factor (BNF), and haemodynamic variables in patients with heart failure. SETTING--University teaching hospital. METHODS--20 patients with impaired left ventricular systolic function [ejection fraction 32 (SEM 2.3)%] were randomised in a double blind manner to receive either oral metoprolol 6.25 mg twice daily or celiprolol 25 mg daily. Haemodynamic variables were evaluated by Swan-Ganz pulmonary artery catheter over 24 hours. ANF and BNF concentrations were measured at baseline, 5 h, and 24 h by radioimmunoassay. RESULTS--At baseline ANF and BNF concentrations were considerably raised compared to the normal range. Treatment with metoprolol caused ANF to rise further to 147% of the basal level at 5 h (P = 0.017) and 112% at 24 h (P = 0.029). This was associated with a small but non-significant rise in pulmonary capillary wedge pressure. Cardiac output and systemic vascular resistance were unchanged at 24 h. In contrast, after celiprolol ANF fell to 90% of basal levels at 5 h and to 74% of basal level at 24 h (P = 0.019), associated with a small but non-significant fall in pulmonary capillary wedge pressure [-3.3 (2.7) mm Hg] and systemic vascular resistance, and rise in cardiac output from 3.2 (0.2) to 4.0 (0.4) l/min (P = 0.04). BNF concentrations rose to 112% of baseline at 5 h (P = 0.09) after metoprolol but fell slightly, to 91% of baseline values, after celiprolol (NS). CONCLUSIONS--Metoprolol, even in very low doses (6.25 mg), produced a rise in ANF and BNF, although minimal haemodynamic changes were detected. In contrast, a vasodilating beta blocker was associated with a significant fall in ANF and BNF and a small rise in cardiac output. This study confirms both the advantages of vasodilating beta blockers over metoprolol for initial treatment of heart failure and the usefulness of ANF and BNF measurements for the assessment of drug effects in heart failure compared to traditional haemodynamic measurements.