PURPOSE: We performed a retrospective review to determine predictive and prognostic factors in patients with metastatic breast cancer who received induction therapy, and, if they responded to treatment, high-dose chemotherapy. PATIENTS AND METHODS: Patients with metastatic breast cancer received induction therapy with doxorubicin, fluorouracil, and methotrexate (AFM). Partial responders then received immediate high-dose chemotherapy, whereas those who achieved complete remission were randomized to immediate or delayed high-dose chemotherapy with hematopoietic stem-cell support. We performed a retrospective review of data from these patients and used Cox proportional hazards regression models for analyses. RESULTS: The overall response rate for the 425 patients enrolled was 74% (95% confidence interval, 70% to 78%). Multivariate analysis of data from all 425 patients revealed that positive estrogen receptor status (P =.0041), smaller metastatic foci ( 2 cm) (P =. 0165), a longer disease-free interval from initial diagnosis to diagnosis of metastases ( 2 years) (P =.0051), and prior treatment with tamoxifen (P =.0152) were good prognostic signs for overall survival. Patients who had received prior adjuvant therapy (P =.0001) and those who developed liver metastases (P =.0001) had decreased long-term survival. In the subgroup of responders to AFM induction, multivariate analysis showed that those with visceral metastases did less well (P =.0006), as did patients who had received prior adjuvant therapy (P =.0023). However, those who had received tamoxifen therapy in the adjuvant setting did better (P =. 0143). CONCLUSION: The chance for long-term remission with induction therapy with AFM and high-dose chemotherapy is increased for hormone receptor positive-patients with nonvisceral metastases who have not received prior adjuvant chemotherapy and have long disease-free intervals.