Retinoic acid (RA) blocks insulin-like growth factor-I (IGF-I) stimulation of proliferation in the MCF-7 breast carcinoma cell line, and this is associated with the appearance of 42- to 46-kilodalton (kDa) IGF-binding proteins(s) (IGFBPs) in the conditioned medium (CM), in addition to the approximately 34- and 27-kDa IGFBPs present in the CM of unstimulated cells. Using immunological, biochemical, and molecular biological criteria, we have identified the 27-kDa band as IGFBP-4, the 34-kDa band as IGFBP-2, and the 42- to 46-kDa band as IGFBP-3. IGF-I alone stimulated MCF-7 cell proliferation, and this was associated with a large increase in IGFBP-2 in the CM. RA alone resulted in increased IGFBP-4 levels and the appearance of IGFBP-3 in the CM. The combination of RA and IGF-I, which resulted in decreased cellular proliferation, was associated with the appearance of IGFBP-3 in the CM at levels far exceeding those seen with RA alone. The effect of IGF-I on IGFBP-2 levels and the synergistic action of IGF-I and RA on IGFBP-3 levels in CM were blocked by alpha IR3, a monoclonal antibody to the human IGF-I receptor, indicating that these effects required signal transduction through the IGF-I receptor. IGFBP-2, -3, and -4 mRNAs were detected in unstimulated MCF-7 cells. RA increased IGFBP-3 mRNA levels, suggesting that transcriptional events contribute to the RA stimulation of IGFBP-3 appearance in CM. In contrast, the increase in IGFBP-2 protein in CM after IGF-I treatment appeared to be greater than the increase in IGFBP-2 mRNA levels. The increase in IGFBP-3 protein in CM in response to the combination of RA and IGF-I was much greater than the increase in IGFBP-3 mRNA. These results suggest that the action of RA and IGF-I in combination to increase IGFBP-3 protein in CM is principally translational or posttranslational. We speculate that RA inhibition of IGF-I-stimulated MCF-7 cell proliferation may be due to IGFBP-3, or that increased levels of IGFBP-3 in response to growth inhibition represent a compensatory response.