Biologic evidence suggests substantial effect of insulin‐like growth factor (IGF)‐I in mammary cell carcinogenesis. However, controversy remains regarding the association between circulating IGF‐I levels and the risk of premenopausal breast cancer in epidemiologic studies. In addition, the association of IGF‐binding protein (IGFBP)‐3, which binds with and modifies the effect of IGF‐I, is unclear. To clarify these associations, we performed a meta‐analysis of all the published studies. A systematic review of literature was conducted. Eligible study designs were nested case‐control and population‐based case‐control studies that give estimates for menopausal women. The studies published between January 1990 and March 2003 were obtained from Medline. We obtained 7 studies, consisting of 688 premenopausal incident breast cancer cases and 1,366 controls, for our final evaluation. Summary statistics were odds ratios (ORs) comparing the highest and the lowest levels of IGF‐I and IGFBP‐3 adjusted for confounders other than IGF‐I or IGFBP‐3. There was neither evidence of heterogeneity between studies nor evidence of publication bias. The confounders considered and the contrast used for the ORs were the major source of variation. The subjects with higher circulating levels of IGF‐I had marginally significant increased risk of breast cancer with an OR of 1.74 (95% CI = 0.97–3.13; p = 0.06). No significant difference was observed for IGFBP‐3 group (OR = 1.60; 95% CI = 0.84–3.02; p = 0.15). In conclusion, we found a marginally significant association between circulating IGF‐I levels and the risk of premenopausal breast cancer. © 2004 Wiley‐Liss, Inc.