A single cycle of high-dose chemotherapy with stem cell support (HDC) in women with responsive metastatic breast cancer (BC) consistently achieves over 50% complete and near complete response (CR/nCR). This significant cytoreduction results in a median event-free survival (EFS) of 8 months, and approximately 20% 3-year and 16% 5-year EFS in selected patients. To improve long-term outcomes, new strategies to treat minimal residual tumor burden are needed. Increased total dose delivered can be achieved with two cycles of HDC. Critical design issues include shortening induction chemotherapy to avoid development of drug resistance and the use of different agents for each HDC cycle. We have determined the maximum tolerated dose (MTD) for paclitaxel combined with high-dose melphalan in the context of a double transplant and explored the impact of a short induction phase. Between June 1994 and August 1996, we enrolled 32 women with metastatic BC on to this phase I double transplant trial. Induction consisted of doxorubicin 30 mg/m2/day days 1-3 given for 2 cycles every 14 days with G-CSF 5 microg/kg s.c. days 4-12. Stem cell collection was performed by leukapheresis in each cycle when the WBC recovered to above 1000/microl. Patients with stable disease or better response to induction were eligible to proceed with HDC. HDC regimen I (TxM) included paclitaxel with dose escalation from 0 to 300 mg/m2 given on day 1 and melphalan 180 mg/m2 in two divided doses given on day 3. HDC regimen II was CTCb (cyclophosphamide 6 g/m2, thiotepa 500 mg/m2, and carboplatin 800 mg/m2 total doses) delivered by 96-h continuous infusion. At the first dose level of 150 mg/m2 paclitaxel by 3 h infusion, four of five patients developed dose-limiting toxicity consisting of diffuse skin erythema and capillary leak syndrome. Only two of these five completed the second transplant. Subsequently, paclitaxel was delivered by 24-h continuous infusion together with 96 h of dexamethasone and histamine receptor blockade. This particular toxicity was not observed again. No toxic deaths occurred and dose-limiting toxicity was not encountered. Three patients were removed from study prior to transplant: one for insurance refusal and two for disease progression. All others completed both cycles of transplant. Complete and near complete response (CR/nCR) after completion of therapy was achieved in 23 (72%) of 32 patients. The median EFS is 26 months. The median overall survival has not yet been reached. At a median follow-up of 58 months, EFS and overall survival are 41% and 53%, respectively. This double transplant approach is feasible, safe, and tolerable. Treatment duration is only 14 weeks and eliminates lengthy induction chemotherapy. The observed event-free and overall survivals are promising and are better than expected following a single transplant. Whilst selection biases may in part contribute to this effect, a much larger phase II double transplant trial is warranted in preparation for a potential randomized comparison of standard therapy vs single vs double transplant.