OBJECTIVE To evaluate characteristics of patients with central nervous system (CNS) lesions of gestational trophoblastic disease (GTD) and determine prognostic and therapeutic implications applicable to management. METHODS We retrospectively reviewed the records of 454 patients treated at the Southeastern Regional Trophoblastic Disease Center between 1966 and 1992 with at least 2 years of follow-up, and identified 42 (9.3%) with CNS metastases. Sixteen patients presented for primary therapy and 27 patients had received significant therapy prior to presentation. Three heavily treated moribund patients died before their first cycle of chemotherapy and were excluded from analysis. Brain metastases were documented by physical exam and radionuclide imaging (before 1976), computed tomography scan (after 1976), or magnetic resonance imaging (after 1986). Patients received multiagent chemotherapy with methotrexate, actinomycin D, and chlorambucil (MAC)- or etoposide-based regimens. All patients received radiation therapy. No intrathecal chemotherapy was given. Craniotomy was employed in seven cases. Remission was defined as three weekly hCG levels below assay sensitivity (< 5 mIU/ml). RESULTS Overall survival was 44%. Twelve of 16 patients (75%) who presented with CNS metastases with no prior therapy (Group A), 5 of 13 (38%) patients who had prior treatment (Group B), and none of 10 patients who developed CNS metastases during therapy (Group C) survived (P < 0.05). Two of four patients who failed in the CNS after treatment for CNS lesions were salvaged. Demographic characteristics of Groups A and B were similar. No significant differences with respect to WHO score, interval from pregnancy to onset of disease, or age among these groups were found. Group B patients had a four-fold higher incidence of liver metastases. Survival of Group A patients was not related to conventional clinical prognostic factors. Inverse (nonsignificant) correlations were found for Group B patients between survival and WHO score, hCG level, size and number of metastatic lesions, but not type of prior therapy. Survival was higher in those with prior molar pregnancies (56%) as contrasted with aborted (50%) or term (27%) gestations. Selective use of craniotomy helped alleviate intracranial pressure and resect refractory foci. CONCLUSIONS Chemotherapy combined with radiation therapy in GTD patients with CNS metastases yields survival rates comparable to those reported for intrathecal methotrexate regimens. Tumor burden as indicated by hCG level and size/number of metastases in previously treated patients may correlate with survival. Patients who develop CNS metastases during active therapy have a very poor outcome.