Circulating levels of insulin-like growth factor binding protein-3 (IGFBP-3) vary greatly between normal individuals, but until now little attention has been given to the study of the genetic factors involved in IGFBP-3 variability in healthy populations. The present study investigated the extent and pattern of the possible genetic influences on plasma levels of IGFBP-3 in 91 nuclear and more complex families, totaling 396 individuals (201 males and 195 females) of Caucasian ethnic origin. The variance decomposition analysis, was performed using the FISHER statistical package. In the second stage of the analysis, we used complex segregation analysis as implemented in the statistical package MAN. Significant negative correlation was revealed between age and plasma levels of IGFBP-3 in both sexes ( r=-0.49; r=-0.23; P<0.001). Multivariate analysis identified age, body weight, and height as significant covariates in men, but for women only age had a considerable effect. It has been demonstrated that about 57.7% of IGBP-3 variation adjusted for significant confounding factors was attributable to genetic factors. The results of bivariate variance decomposition analysis showed no significant genetic and phenotypic correlation between the mineral density of hand bones and IGFBP-3. Segregation analysis revealed the existence of a potential major gene effect that was able to explain some 27.5% of IGFBP-3 variation. Multifactorial effects, likely, unknown minor genes, contributed an additional 30% to IGFBP-3 variation. The segregation analysis also provided evidence of significant genotype X sex interaction in the determination of plasma levels of IGFBP-3.