The human tumor clonogenic assay (HTCA) was evaluated in 407 fresh samples of breast cancer from 288 patients. Seventy samples were inadequate for testing. Adequate in vitro growth for drug testing (greater than 30 colonies/plate) was obtained in 91 (27%) of the 337 viable samples, inadequate growth for drug evaluation (5 to 30 colonies/plate) in 17%, and no colony formation (less than 5 colonies/plate) in 56%. Operationally defining a greater than or equal to 50% inhibition of colony formation as in vitro drug sensitivity, the in vitro response rates to 12 anticancer drugs tested against ten to 36 different cancers (arranged in decreasing order according to the number of tests performed) were as follows: doxorubicin (14%), bisantrene (54%), vinblastine (33%), mitomycin (36%), interferon clone A (23%), 5-fluorouracil (20%), methotrexate (17%), leukocyte interferon (33%), mitoxantrone (42%), cyclophosphamide (25%), m-AMSA (16%), and melphalan (10%). Among 25 patients receiving single-agent therapy, there were ten (59%) of 17 with in vitro sensitivity who responded; resistance was correctly predicted in nine patients (100%), P = .01. Among 34 patients treated with combination chemotherapy, seven (50%) of 14 with in vitro sensitivity responded, and resistance was predicted in 13 (65%) of 20 patients. Difficulties in using the HTCA in breast cancer (including small specimen size, difficulties in disaggregation, and inadequacy of growth) will require additional research. Nonetheless, the assay appears to detect in vitro activity as well as resistance of a variety of anticancer agents and appears to predict clinical responsiveness to standard as well as some investigational single agents.