87 nonpretreated stage III/IV ovarian common epithelial carcinomas were studied for estrogen receptor (ER) and progesterone receptor (PR) content by both immunohistochemistry (IHC) and biochemical (DCC) analysis. While the DCC assay showed tumors to be receptor-positive in 62% (ER) and 66% (PR), receptor-positive malignant epithelial cells were only detected in 38% (ER) and 31% (PR) by IHC. There was only a low correlation between the semiquantitative results of ER and PR IHC and the corresponding values of DCC receptor determination. The finding of steroid receptor-positive stromal cells without any evidence of hormone receptor-positive epithelial tumor cells offers a possible explanation for discrepant results in numerous cases with obviously "false positive" results of DCC analysis. Since the considerable heterogeneity of steroid receptor expression present in many ovarian neoplasms can only be detected by IHC, it seems to be the appropriate method of ER and PR determination. Most patients were treated by both radical cytoreductive surgery (n = 76) and a platinum-based chemotherapy (n = 79). ER was not shown to be of significant prognostic value. However, survival was significantly better in patients with PR positive tumors (IHC and DCC) on univariate analysis. Residual tumor after primary surgery was the only remaining significant prognostic factor after multivariate analysis. Further studies are needed to clarify the biological function of steroid receptor-positive stromal cells in ovarian carcinomas.