The purpose of this study was to investigate the treatable subsets in cancer of unknown primary origin (CUP). Fifty patients (27 males and 23 females; median age, 53 years) with CUP diagnosed between April 1992 and June 1999 were analyzed retrospectively. Of the 50 patients, 39 received chemotherapy: platinum-based in 31, non-platinum-based in 4, and clinical trials of new agents in 4. Of the 39 patients, 13 (33.3%; 95% confidence interval: 19.1-50.2%) showed objective responses, with 4 complete responders. Patients with poorly differentiated carcinomas in whom β β β β-subunit of human chorionic gonadotropin (β β β β-HCG) was elevated more than 10 mIU/ml and female patients with peritoneal adenocarcinomatosis achieved high response rates (83.3% and 80%, respectively) with platinum-based chemotherapy, as compared with only a 15.3% response rate in the remaining patients. Platinum-based chemotherapy provided promising results in patients with poorly differentiated carcinomas and in female patients with peritoneal adenocarcinomatosis. Significantly elevated serum levels of β β β β-HCG in patients with poorly differentiated carcinoma might predict a better response to platinum-based chemotherapy. However, the investigation of novel chemotherapeutic approaches is warranted for other groups of patients with CUP.Key words: Cancer of unknown primary -Chemotherapy -Chemosensitive subgroups -β-HCGThe development of imaging procedures, such as ultrasonography, computed tomography, and magnetic resonance imaging, has enabled the detection of the site of primary cancers, as well as metastatic sites. Moreover, immunohistochemical studies were introduced into routine clinical oncology practice recently and have significantly increased the probability of identifying the likely underlying tumor type.1) In spite of these notable advances, the primary site of cancer can still not be determined in approximately 5-10% of all cancer patients.2) As a result, these patients are diagnosed as having "cancer of unknown primary" site (CUP). Accordingly, CUP, by its nature, is extremely heterogeneous in clinical presentation, histologic appearance, and natural history, and there is no standard treatment for CUP. Although reports of patients with CUP have indicated poor prognosis, with overall survival duration of 5-11 months from diagnosis of CUP, [3][4][5] several favorable prognostic factors for chemotherapeutic response have been proposed and the treatment results of patients with such factors were markedly better. 6) However, there is very little information concerning Japanese patients with CUP.To establish such treatable subsets and to identify further prognostic factors predictive of positive responses to chemotherapy, Japanese patients with CUP were retrospectively analyzed after appropriate clinical and pathological exclusion of primary tumors, according to current diagnostic recommendations.