Meningiomas often contain steroid hormone receptors, but the correlation of receptor presence with patient outcome or mitotic index is unclear. Intracranial meningiomas from 70 patients (27 males and 43 females, mean age 52.9 + 1.7 years [mean +/- standard error of the mean], range 15-78 years) were evaluated immunocytochemically for female sex hormone receptors using specific monoclonal antibodies. Prognostic correlations were determined using statistical analyses that included clinical and histological variables. Twenty-eight tumors were benign, 27 had atypical features, and 15 were malignant. Thirty tumors were meningotheliomatous, 11 were fibroblastic, 28 were transitional, and one was secretory. Twenty-nine of the 70 primary tumors recurred (mean interval to recurrence 50.1 +/- 10 months). The mean progression-free follow-up period for patients without recurrence was 82.1 +/- 7.7 months. Nuclear staining for the progesterone receptor (PR) was found in 58 cases (83%) and PR status was scored as 0 (0% nuclei positive), 1 (< 1%), 2 (1-9%), 3 (10-49%), or 4 (> 50%). Only six tumors (8.6%) contained nuclear estrogen receptor (ER) staining, which was limited to a small number of nuclei (< 1%). Fisher's exact test (two-tailed) showed an inverse correlation between tumor grade and PR staining score (p < or = 0.001), with 96% of benign and 40% of malignant meningiomas containing PR-positive nuclei. No correlation between age or histological subtype and PR score was detected. Meningiomas from female patients had more PRs (p < or = 0.05). Analysis of variance revealed that the mitotic index (total counts of mitoses per 10 high-power fields) for tumors with 0 PR staining (18 +/- 4.4) was higher (p < or = 0.0001) than for those with PR scores of 1 to 4 (4.3 +/- 1.9, 5.1 +/- 2, 2.2 +/- 0.8, and 1.7 +/- 0.9, respectively). Univariate analysis indicated that the absence of PR, high mitotic index, and higher tumor grade were significant factors for shorter disease-free intervals. Multivariate analysis showed that a three-factor interaction model, with a PR score of 0, mitotic index greater than 6, and malignant tumor grade, was a highly significant predictor (p < or = 0.0001) for worse outcome in patients harboring meningiomas. These data indicate that the presence of PRs, even in a small number of tumor cells, is a favorable prognostic factor for meningiomas.