We have explored several novel high-dose combinations in an attempt to increase antitumor activity while decreasing treatment-related toxicity. From October 1989 through June 1997, we performed phase I/II dose-escalation trials exploring novel high-dose regimens including ifosfamide/carboplatin/etoposide, mitoxantrone/thiotepa, and paclitaxel (Taxol; Bristol-Myers Squibb Company, Princeton, NJ)/mitoxantrone/thiotepa. We have also evaluated busulfan/cyclophosphamide and cyclophosphamide/thiotepa/carboplatin in phase II trials. Three hundred ninety-three patients have been treated in these trials and followed for a minimum of 3 months. Event-free survival (including relapses and treatment-related mortality; +/-SE) at 3 years by stage and chemosensitivity is as follows: stage II, four to nine positive nodes (n=16), 52%+/-17%; stage II, greater than nine nodes (n=30), 46%+/-11%; stage III (n=59), 50%+/-8%; inflammatory stage III (n=15), 27%+/-17%; stage IV, anthracycline responsive (n=69), 19%+/-5%; stage IV, anthracycline refractory but responsive to salvage therapy with ifosfamide, carboplatin, and etoposide or paclitaxel (n=53), 12%+/-6%; stage IV, refractory (n=128), 5%+/-2%; and stage IV, not evaluable for response (n=23), 10%+/-8%. Treatment-related mortality was 4% for both phase I and II studies involving stage II breast cancer patients, 5% for stage III breast cancer, 15% for inflammatory breast cancer, and 18% for all stage IV breast cancers, responsive and refractory. We conclude that high-dose therapy for the treatment of high-risk early stage breast cancer or metastatic breast cancer results in durable remissions. Chemosensitivity to induction regimens remains the most important prognostic indicator, although long-term survival has been seen even in patients with highly refractory disease. Further studies are necessary to define optimal high-dose strategies based on stage and chemosensitivity of disease.