We studied the relationships in 29 patients with cirrhosis between pulmonary arterial atrial natriuretic peptide concentrations and the following: systemic and splanchnic hemodynamics, the hematocrit, arterial oxyhemoglobin saturation, oxygen tension and the severity of cirrhosis. Plasma atrial natriuretic peptide concentrations ranged from 21 to 208 pg/ml and averaged 78 +/- 8 pg/ml (mean +/- S.E.M.). Simple regression analysis showed significant correlations between plasma atrial natriuretic peptide concentration and the following: hematocrit, mean pulmonary arterial pressure, wedged hepatic venous pressure, free hepatic venous pressure, pulmonary wedged pressure and serum bilirubin concentrations. No significant correlations were found between plasma atrial natriuretic peptide concentrations and all other hemodynamic values, arterial oxyhemoglobin saturation and oxygen tension. Multiple stepwise regression analysis showed that the hematocrit, mean pulmonary arterial pressure and wedged hepatic venous pressure were significant and independent predictors of pulmonary artery plasma atrial natriuretic peptide concentrations (R2 = 0.69). Partial regression coefficients were -0.74 (p less than 0.001), 0.61 (p less than 0.001) and 0.44 (p less than 0.05) for the hematocrit, the mean pulmonary arterial pressure and the wedged hepatic venous pressure, respectively. In conclusion, in patients with cirrhosis, increased plasma atrial natriuretic peptide concentrations were related to the degree of hemodilution, increased pulmonary arterial pressure and the degree of portal hypertension. Plasma atrial natriuretic peptide concentrations were not influenced by the arterial oxygenation levels.