Malignant gliomas--glioblastoma multiforme and anaplastic astrocytoma--are among the most fatal forms of cancer in humans. It has been suggested that hepatocyte growth factor (HGF) is a reliable predictor of glioma malignancy; amounts of HGF are directly related to cellular proliferation, angiogenesis, low apoptotic rate, and poor prognosis (WHO III and IV). We measured the HGF content of cerebrospinal fluid (CSF) from patients with malignant glioma glioblastoma multiforme (WHO IV; n = 14), anaplastic astrocytoma (WHO III; n = 4), and meningioma (WHO I; n = 9), and from control subjects (n = 25), and found a high concentration of HGF in patients with malignant glioma. However, CSF concentrations from glioblastoma multiforme and anaplastic astrocytoma patients were not statistically significantly different (893 +/- 157 vs. 728 +/- 61, respectively; P > 0.01). A negative correlation between HGF and survival was found at five years of follow-up (R = -0.922, R (2) = 0.850, P < 0.001). Also, the HGF concentration in CSF was a reliable means of explaining the highly variable survival of patients with malignant glioma. CSF concentrations of HGF higher than 500 pg/ml were associated with increased mortality whereas values higher than 850 pg/ml were associated with a brief tumor-free period after surgery (9 +/- 0.6 vs. 6 +/- 0.6 months, respectively, P < 0.001). Our findings support the idea that measurement of HGF in CSF could be a useful tool for monitoring the biological activity of malignant glioma. The findings will ultimately need to be confirmed in a much larger study.