Background The mechanism of atrial natriuretic peptide (ANP) release has been difficult to demonstrate in patient studies because of inaccuracies in measuring atrial volumes using conventional techniques. Methods Magnetic resonance imaging was performed in 28 clinically stable patients (New York Heart Association class 3) with chronic heart failure to determine right atrial (RA), left atrial (LA), and ventricular volumes. In addition, right heart catheterization was serially performed and plasma ANP levels (in picograms per milliliter) were drawn from the right atrium. Results Five patients had to be excluded from data analysis for technical reasons. The remaining 23 patients had the following hemodynamic measurements (mean ± SD): RA mean pressure 7 ± 5 mm Hg, pulmonary artery mean pressure 28 ± 10, pulmonary capillary wedge pressure 21 ± 8 mm Hg, and cardiac index 2.9 ± 1.4 (L/min/m2), respectively. Plasma ANP levels were significantly elevated at 162 ± 117 (normal range 20 to 65 pg/ml, p < 0.05), as were LA and RA volumes compared with healthy controls (RA volume 128 ± 64 ml vs 82 ± 25 ml, p < 0.05; LA volume 157 ± 54 ml vs 71 ± 24 ml, p < 0.01, respectively). ANP showed a stronger relation with atrial volumes (RA volume, r = 0.91, p = 0.0001; LA volume, r = 0.80, p = 0.001) than with atrial pressures (RA mean pressure, r = 0.45, p = 0.03; pulmonary capillary wedge pressure, r = 0.67, p = 0.001). A subgroup analysis of patients with increased RA or LA volumes (>1 SD of mean of controls) revealed a stronger relation between ANP and RA volumes than between ANP and LA volumes. Conclusions These data suggest that increased right heart volume with subsequent increased atrial stretch is the major determinant for ANP release in patients with stable CHF. (Am Heart J 1997;135:592-7.)