Introduction: We examined the relationships between insulin-like growth factor I (IGF-I), its binding protein (ICFBP-3), body composition, and bone mineral density (BMD) in collegiate runners (N = 13), gymnasts (N = 10), and noncompetitive women (N = 10). Methods: Subjects were evaluated by dual-energy x-ray absorptiometry for body composition and BMD of the spine, hip, and whole body, fasting serum levels of ICE-I and IGFBP-3, and dietary intake. The ratio IGF-I/IGFBP-3 was calculated as a marker of IGF-I bioavailability. Results: In ANOVA, IGF-I and IGF-I/IGFBP-3 in athletes with oligomenorrhea and amenorrhea did not differ from eumenorrheic athletes; thus, values were pooled. Lean/height(2) and bone mass at the hip and spine were higher in gymnasts than runners and controls. Total caloric intake was similar between groups. ICE-I and IGF-I/IGFBP-3 differed between groups with gymnasts having higher IGF-I values than runners (397 +/- 58 vs 288 +/- 73 ng(.)mL(-1), P < 0.001) and higher IGF-I/IGFBP-3 than controls and runners (0.065 +/- 0.009 vs 0.056 +/- 0.008 vs 0.045 +/- 0.009, P = 0.0001). In simple regression, IGF-I and IGF-IGFBP-3 were related to lean/height(2) and BMD of the lumbar spine and hip (P < 0.01-0.0001). IGF-I and IGF-I/IGFBP-3 were multicollinear: thus, the ratio was used in subsequent stepwise regression. Lean mass, corrected for body surface area (height(2)), independently predicted spine and trochanteric BMD (R-2 = 0.26, 0.28, respectively), whereas IGF-I/IGFBP-3 and lean/height(2) together contributed to 48% of the variance in femoral neck BMD. Conclusion: We conclude that, in this group Of young adult women, lower BMD in runners may be due, in part, to lower levels of IGF-I and the ratio of IGF-I-to-IGFBP-3 and that IGF-I may mediate the relationship between bone and lean mass.