Measurements of human chorionic gonadotrophin (HCG) concentration i n plasma and cerebrospinal fluid (CSF) have been made on patients with gonadotrophin-producing tumors. I n the absence of brain metastases the spinal fluid concentration is, within wide limits, proportional to that in the plasma. In 73 patients with gestational choriocarcinoma the mean plasma/spinal fluid ratio was 286 with a lowest value of 64. In five patients with gonadotrophin-producing teratomas the mean ratio was 208 with a lowest value of 104. In 29/33 patients with brain metastases the p1asma:CSF ratio was less than 60 at the time confirmatory evidence of brain metastases was obtained. One patient with a gonadotrophin-producing teratoma h a d a brain metastasis that apparently failed to produce HCG and this metastasis failed to show the histologic features of choriocarcinoma. Monitoring the plasma : CSF ratio provided evidence of brain metastases in 13/18 patients who were undergoing chemotherapy for extensive metastatic disease before confirmatory evidence was obtained by other methods. T h e lead-in time between assay diagnosis and diagnosis by other methods ranged from 1-20 weeks. Monitoring the p1asma:CSF ratio provided a means of observing the response of cerebral metastases to therapy. I n some patients the CSF HCG concentration exceeded the plasma concentration indicating that the higher CSF values in patients with CNS metastases cannot be attributed to impairment of the blood-brain barrier. Direct secretion of tumor products into CSF o r indirect secretion into CSF via cerebral extracellular fluid evidently occurs. I n contrast with radiographic a n d radionucleide detection methods, a chemical marker indicates the metabolic activity of tumor cells within the central nervous system and provides a basis for monitoring that activity.