We measured human chorionic gonadotropin by 2 immunometric assays that require that the entire human chorionic gonadotropin molecule is intact, and by a competitive radioimmunoassay that measures intact human chorionic gonadotropin and its free beta-chain. The sera tested were samples from male cancer patients with elevated human chorionic gonadotropin levels obtained by the radioimmunoassay method. Elevated levels were confirmed in 67 of 92 samples (72 per cent) with the immunometric methods. However, in 25 of 97 patients (28 per cent) elevated human chorionic gonadotropin was found with the radioimmunoassay, whereas the values were in the normal range when measured with the immunometric assays. These discrepancies were found in 22 patients with seminomatous tumors, including 2 extragonadal germ cell tumors, which constitutes 42 per cent of all seminoma patients tested. Of the remaining 3 discrepant patients 2 had lung cancer and 1 had metastases from an unknown primary cancer.