The aim of this study was to perform a multivariate analysis including clinical and biological prognostic factors on glial tumor outcome. Seventy-nine patients were analyzed (48 men and 31 women; mean age = 56 years, range = 16-77 years): 7 had a benign glial tumor (grades 1 and 2), 21 had an anaplastic glial tumor (grade 3), and 51 had a glioblastoma (grade 4). Median follow-up was 17.9 months for patients who survived (50 patients died). Biopsies were obtained at time of diagnosis (complete tumor resection in 62 patients and stereotaxic biopsies in 17 patients). Epidermal growth factor receptor (EGFR) was measured by a binding assay, and labeling index (LI) was measured by tritiated thymidine incorporation. EGFR varied from 4 to 73,110 fmol/mg protein (mean = 3912 fmol/mg protein; median = 374 fmol/mg protein; n = 79). LI varied between 0.1 and 16.5% (mean = 6.2%; median = 5.2%; n = 40). Log10 EGFR was significantly and positively correlated with patient age. LI was significantly different according to tumor histology. Univariate Cox analysis (end point was cancer death) showed that age (P = 0.027), log10 EGFR (P = 0.025), and LI (P = 0.0019) were significant continuous variables, the survival being shortened when the covariable increased; tumor resection (P = 0.015, relative risk = 0.45) and histology (P = 0.0009) were significant categorical factors. A multivariate Cox analysis (forward selection) including age, histology, tumor resection, log10 EGFR, and LI revealed that log10 EGFR, LI, and tumor resection were the only independent significant predictors of survival. This multivariate approach reveals that the clinical prognostic factors of glial tumors, namely age and tumor histology, disappear, to the benefit of intrinsic characteristics of the tumor, i.e., EGFR expression and LI, suggesting that coupled EGFR and LI determination could be a useful tool for better evaluation of glial tumor outcome.