We have reported that paclitaxel results in cisplatin sensitization in cisplatin-resistant ovarian cancer cell lines in vitro and in nude mice. The purpose of this trial was to determine the maximum tolerated dose and recommend phase II dose of weekly single agent paclitaxel for outpatients with recurrent or persistent epithelial ovarian carcinoma (REOC), with standard chemotherapy containing platinum in the initial setting. Patients with REOC were eligible for this protocol regardless of the number and kind of previous chemotherapy regimens. The starting dose was paclitaxel 70 mg/m2/week in 1-hour infusion, 3 weeks on, 1 off and repeated at least twice. This dose was increased by 10 mg per step to 100 mg/m2/week. Three patients were accrued to each dose cohort. Three new patients were to be entered at escalation doses unless dose-limiting toxicities (DLT) occurred, defined as grade 4 hematological or grade 3/4 non-hematological toxicities. If 1 out of 3 patients developed DLT, 3 additional patients were entered at the same dose level. Sixteen patients were accrued. All the patients had received at least one prior platinum-containing regimen (1 regimen 14 cases, 2 regimens 1 case, 3 regimens 1 case). At the level I dose of 70 mg/m2/week no hematological or non-hematological toxicity more than grade 2 was observed. At the level II dose of 80 mg/m2/week, 1 patient had grade 4 non-hematological toxicity, showing difficulty-walking. Three new additional patients were treated with the same dose. Except for this patient, 1 had grade 3 leukopenia and grade 4 neutropenia, but these toxicities were overcome within 3 days without support of granulocyte-colony stimulating factor (G-CSF). At the level III dose, 90 mg/m2/week, 1 of 3 patients showed grade 4 leukopenia and 2 had grade 4 neutropenia, requiring support by G-CSF. Similarly, when using 100 mg/m2/week of paclitaxel, 2 out of 4 patients had more than grade 3 hematological toxicity. However, at levels II or IV, no non-hematological toxicity exceeding grade 2 was observed. Even if the weekly single paclitaxel was repeated, the toxicity did not seem to accumulate. According to dose-escalation, use of G-CSF and treatment delay were increased. The use of G-CSF was significantly (p<0.05) increased between levels I, II, III and IV. Although treatment with 90 or 100 mg/m2/week, at 3 weeks on, 1 off was tolerable and safe with support of G-CSF, these doses cannot be recommended for out-patients because of treatment delay. In this phase I trial, 80 mg/m2/week of paclitaxel was recommended as the phase II dose for outpatients.