We recently reported on higher expression of the insulin-like growth factor pathway genes IGF-II and IGFBP3 in serous ovarian/peritoneal carcinoma compared to malignant peritoneal mesothelioma. The present study analyzed the diagnostic and clinical role of these proteins in serous effusions. Effusions (n = 327), including 294 carcinomas (205 ovarian, 48 breast, 17 cervical/endometrial, 12 lung, 12 gastrointestinal/genitourinary) and 33 malignant mesotheliomas, were immunostained for insulin-like growth factor-II and insulin-like growth factor binding protein-3. Surgical ovarian carcinoma (n = 124) and peritoneal mesothelioma (n = 18) specimens were additionally studied. Insulin-like growth factor binding protein-3 levels were measured in 148 effusion supernatants (114 ovarian carcinomas, 18 breast carcinomas, 16 mesotheliomas) using enzyme-linked immunosorbent assay. Insulin-like growth factor binding protein-3 promoter methylation was analyzed in 11 ovarian carcinoma effusions. Insulin-like growth factor binding protein-3 (P = .002) and insulin-like growth factor-II (P < .001) expression by immunohistochemistry was significantly higher in carcinomas compared to mesotheliomas, with diagnostic sensitivity of 77% and 70% and specificity of 55% and 70%, respectively. In surgical specimens, insulin-like growth factor binding protein-3 expression was higher in ovarian carcinomas compared to peritoneal mesotheliomas (P = .007), whereas insulin-like growth factor-II expression was comparable (P = .505). Insulin-like growth factor binding protein-3 levels by enzyme-linked immunosorbent assay were comparable in the 3 analyzed cancer types. Insulin-like growth factor binding protein-3 promoter methylation was found in 6 of 11 effusions. High insulin-like growth factor binding protein-3 expression in prechemotherapy and high insulin-like growth factor-II expression in postchemotherapy ovarian carcinoma effusions correlated with poor overall survival (P = .031 and P = .024, respectively). Insulin-like growth factor-II expression in postchemotherapy effusions was an independent prognostic factor in Cox multivariate analysis (P = .04). In conclusion, insulin-like growth factor-II and insulin-like growth factor binding protein-3 are more frequently expressed in metastatic carcinomas compared to mesothelioma in effusions but are less specific than currently used markers. Insulin-like growth factor-II and insulin-like growth factor binding protein-3 may be novel prognostic markers in metastatic ovarian carcinoma.