We examined the records of 82 patients with clinical Stage I nonseminomatous germ cell tumors of the testis who, after radical orchiectomy, were treated by surveillance at M.D. Anderson Cancer Center between October, 1981, and March, 1987. Our purpose was to determine whether or not patients with a low risk of relapse can be identified at the time of the initial staging evaluation. In 30 of 82 patients (Group 1), embryonal carcinoma constituted less than 80 percent of the tumor, no vessel invasion was present, and the preorchiectomy serum AFP level was less than 80 ng/dL. No relapses occurred in this group. Fifty-two patients (Group 2) had more than 80 percent embryonal carcinoma or vessel invasion or a serum AFP level higher than 80 ng/dL. Relapse occurred in 24 (46%) of these patients. The difference in the rate of relapse between patients in Group 1 and Group 2 was statistically significant (P less than 0.00001). A separate analysis of teratoma as a predictor of nonrelapse showed that the orchiectomy specimens of 30 of the 82 patients contained more than 50 percent teratoma. Only 1 relapse occurred among 25 patients with more than 50 percent teratoma and no vessel invasion. Our data show that there is a subgroup of patients with clinical Stage I nonseminomatous germ cell tumor who have a very low rate of relapse. We believe these patients can be effectively treated by surveillance and should be spared the morbidity of an unnecessary retroperitoneal lymph node dissection.