To clarify the involvement of atrial natriuretic peptide (ANP) in the pathogenesis of liver cirrhosis, we measured plasma ANP in patients with various stages of cirrhosis and in age-matched normal subjects. Urinary cyclic guanosine monophosphate (cGMP) was also measured as a marker of active biological ANP. In addition, effects of exogenous synthetic human ANP (0.5 micrograms/kg) on renal functions were examined in normal subjects and in cirrhotics without ascites or with mild ascites. Plasma ANP levels were not significantly different among these 3 groups. Urinary cGMP concentrations were significantly higher in both cirrhotics without ascites and cirrhotics with mild ascites, (340 pmol/ml, P less than 0.05 and 496 pmol/ml, P less than 0.01 respectively) than normal subjects (95 pmol/ml). In normal subjects, marked increases in urinary volume (UV), sodium excretion (UNaV), fraction excretion of sodium (FENa) and free water clearance (CH2O) were induced after ANP infusion, and significant recoveries were subsequently observed in these parameters. However, in cirrhotics, the responses to ANP infusion of UV, FENa and CH2O were far less dramatic. The response of UV, UNaV and FENa in cirrhotics with mild ascites was delayed compared to cirrhotics without ascites. These results suggest that the blunted natriuretic responsiveness to ANP is contributory to the pathogenesis of initial sodium retention in cirrhotics.