Recent studies suggest that leptin, the ob gene product absent in ob/ob mice, is a negative regulator of adiposity. We developed an RIA to measure human leptin in plasma or serum. The minimum detectable concentration by the assay is 0.5 microg/L leptin and the limit of linearity is 100 microg/L. Recovery of leptin added to serum was 99-104% over by the linear range of the assay. The RIA agreed reasonably well with rough quantification by Western blot (RIA = 0.90 blot + 3.7 microg/L, Sy/x = 10.9 microg/L). CVs within- and between-run ranged from 3.4% to 8.3% and from 3.6% to 6.2%, respectively. Variation in plasma leptin concentrations in specimens collected on consecutive mornings was large (CVs of 10.9% and 22.5%). After an overnight fast, leptin concentrations were similar to those 1-2 h after 1-2 meals. Plasma leptin concentrations in specimens from 83 lean and obese adults correlated directly with body mass index (BMI; kg/m2): r = 0.72, P <0.001. Correlations were significantly improved by separating results by gender (men r = 0.84, women r = 0.87; p <0.001). The increase in leptin concentrations with increasing BMI was greater in women than in men (slope 2.53 vs 0.97 microg/L per unit BMI, respectively). Leptin concentrations determined in lean subjects (BMI between 18 and 25) were higher in women (7.36 +/- 3.73 microg/L) than in men (3.84 +/- 1.79 microg/L) (P <0.001). Plasma leptin varied little with age and no significant difference was observed between whites and blacks. We conclude that: (a) plasma leptin concentrations are accurately and precisely measured by this new RIA; (b) leptin concentrations vary little due to short-term fasting, age, or race; but (c) plasma leptin concentrations are gender specific.