Background: The insulin-like growth factor (IGF)-binding proteins (IGFBPs) regulate the actions of the IGFs by influencing interactions between the IGFs and the IGF receptors. IGFBP-3, one of the six known species of IGFBPs, is the predominant IGFBP in serum and is expressed by breast cancer cells. Compared with estrogen receptor (ER)-positive samples, ER-negative breast cancer cell lines and tumors express higher levels of IGFBP-3. Therefore, expression of IGFBP-3 may be relevant in breast cancer biology, although it is unknown whether IGFBP-3 levels correlate with other breast cancer prognostic factors besides ER status. It is also not known how different methods used to measure IGFBP-3 in breast cancer correlate. Purpose: We measured IGFBP-3 messenger RNA (mRNA) and protein levels in breast tumors by different methods to test how these methods compare and to investigate the relationship between IGFBP-3 and breast cancer prognostic factors. Methods: We analyzed 40 human breast tumors and examined IGFBP-3 expression by ligand blot analysis, immunoblot analysis, immunoradiometric assay (IRMA), and ribonuclease protection assay. Another set of 40 breast tumors, selected according to ER and progesterone receptor (PR) status, S phase, and ploidy, was analyzed by IRMA. Results: In 26 (65%) of 40 samples in which RNA could be isolated, IGFBP-3 mRNA levels correlated with IGFBP-3 levels measured by IRMA (two-sided; P = .0001) but not with IGFBP-3 levels measured by ligand blot or immunoblot. Protein levels were highly correlated among all protein assays. Because the IRMA was more sensitive and accurate than the ligand blot and immunoblot assays, we used IRMA to examine IGFBP-3 levels in an additional 20 primary breast tumors with poor prognostic features (ER and PR negativity, high S phase, and aneuploidy) and in 20 tumors with good prognostic factors (opposite features). IGFBP-3 levels were threefold higher in tumors with poor prognostic features (mean ± standard deviation = 32.8 ± 25.2 versus 11.8 ± 9.7 ng/mg; two-sided; P = .003). Conclusions: These findings suggest that in human breast cancer, IGFBP-3 mRNA and protein levels are correlated and higher levels of IGFBP-3 are detectable in tumors with poor prognostic features. Implications: IGFBP-3 may be involved in the regulation of breast cancer cell growth.