OBJECTIVE To test the hypothesis that variations in serum insulin concentrations and insulin action may influence serum concentrations of lipoprotein(a) [Lp(a)]. METHODS A cross-sectional analysis of fasting serum insulin and Lp(a) concentrations were conducted in a group of 54 healthy adult men 23-61 years of age. Measures of dietary intake, exercise, smoking, alcohol ingestion, body mass index (BMI), fasting plasma lipoprotein lipid concentrations, and serum sex hormone and fasting glucose levels were also determined. RESULTS The fasting serum concentrations of insulin and Lp(a) were negatively correlated (r = -0.339, P = 0.014) by univariate regression analysis. Several confounding variables were also significantly correlated with Lp(a) concentrations: testosterone (r = 0.348, P = 0.012), strenuous exercise (r = 0.287, P = 0.041), and BMI (r = -0.276, P = 0.050). As expected, these additional variables and serum insulin concentrations were highly interrelated. In a stepwise regression model, the serum insulin level was identified as the single best predictor of Lp(a) levels. Nearly 25% of the heterogeneity in serum concentrations of Lp(a) could be predicted by serum levels of insulin and testosterone, BMI, and the amount of strenuous exercise. CONCLUSIONS The results of these studies have shown that serum concentrations of insulin and testosterone, BMI, and strenuous exercise appear to account for the majority of predicted nongenetic variability in serum levels of Lp(a). These observations suggest the possibility in this group of healthy men that serum concentrations of Lp(a) may be modulated by a complex interplay between insulin action, obesity, androgen levels, and strenuous exercise.