The purpose of this study was to evaluate the frequency of detecting occult tumor cells in peripheral blood stem cell (PBSC) harvests and to determine the impact of infusing such cells on relapses after high-dose chemotherapy (HDC). Peripheral blood stem cell harvests from 223 patients with breast cancer were examined by an immunocytochemistry (ICC) method for detection of occult tumor cells, and infused after HDC without consideration of test results. Two hundred and four patients, 114 with stage II-III and 90 with stage IV disease who received only PBSC, that were tested by ICC were evaluated for time to relapse. Five hundred and eighty-one of 619 PBSC harvests (94%) from 223 patients were tested. Fifty-three of 581 harvests (9%), 8% from stage II-III and 10% from stage IV patients, were positive by ICC (P = 0.68). Forty-one of 223 patients (18%), 17/122 (14%) with stage II-III and 24/101 (24%) with stage IV disease, had positive harvests (P = 0.06). Eleven percent of patients who had 1-2 harvests tested were positive as compared to 32% of patients who had > or =3 PBSC harvests tested (P < 0.001). Nineteen patients who were infused with a mixture of ICC negative and untested PBSC harvests were excluded from analyses of relapse. The probabilities of relapse at 18 months for the 97 patients with stage II-III disease infused with ICC-negative and the 17 with ICC-positive PBSC were 0.19 and 0.13, respectively (P = 0.48). The probabilities of relapse at 18 months for patients achieving a CR or a CR in non-bone sites and improvement in bone lesions were 0.55 for the ICC-negative group (n = 30) and 0.45 for the ICC-positive group (n = 11) (P = 0.60). It was concluded that occult tumor cells were detected by ICC in PBSC harvests from a relatively small fraction of women with breast cancer, but were not associated with a significant increase in the probability of early relapse or progression when infused after HDC.