Assays for natriuretic peptides have received considerable attention as potential screening tests for congestive heart failure and left ventricular dysfunction. However, information regarding the impact of age, sex, and other physiologic characteristics on natriuretic peptide levels is limited. We examined a healthy reference sample of 911 subjects (mean age 55 years, 62% women) from the Framingham Heart Study who were free of hypertension, valvular disease, diabetes, atrial fibrillation, obesity, coronary heart disease, congestive heart failure, and renal failure, and who had normal left ventricular systolic function. Plasma brain natriuretic peptide and N-terminal atrial natriuretic peptide levels were measured, and multivariable regression used to assess correlates of natriuretic peptide levels. The strongest predictors of higher natriuretic peptide levels were older age and female sex. Other multivariable predictors included lower diastolic blood pressure (higher pulse pressure), lower body mass index, and higher left atrial size. Reference limits were then formulated based on the empirical distribution of natriuretic peptide levels by gender both across all ages and partitioned by age. Age-pooled reference limits compared with age-specific limits classified a higher proportion of healthy elderly subjects (17% vs 2.5%), but a lower proportion of healthy young subjects (1% vs 2.5%) as "abnormal." We conclude that interpretation of natriuretic peptide levels should take into consideration gender and possibly age. The reference limits derived from this large, healthy community-based sample will aid in the identification of elevated natriuretic peptide levels in clinical practice.