The medical treatment of squamous cell cervical carcinoma is receiving increasing attention. Cisplatin and ifosfamide are known effective drugs. Paclitaxel has been tested with interesting results in cervical cancer. We evaluated the toxic effects and the antitumor activity of a multiagent regimen that included paclitaxel, ifosfamide, and cisplatin (TIP) in two different settings: bulky and locally advanced cervical cancer and recurrent-persistent disease. Treatment consisted of paclitaxel 175 mg/m2 given over 3 hours on day 1, cisplatin 50 mg/m2 (75 mg/m2 in 24 patients), ifosfamide 5 g/m2 and mesna 5 g/m2 given on day 2, and mesna 3 g/m2 given on day 3. In the neoadjuvant setting, the course was repeated every 3 weeks for three courses. Unless there was progression of disease or reason to avoid surgery, all patients were scheduled for radical hysterectomy and pelvic lymphadenectomy. Thirty-eight patients with locally advanced cervical cancer were studied: 11 women achieved a clinical complete response, 21 had a partial response, five had stable disease, and one had disease progression. Among the 34 patients who underwent surgery, six had a pathologically documented complete response, seven had an optimal partial response (only microscopic residual disease), 19 had a suboptimal partial response, and two stable diseases. Grade 3-4 neutropenia was recorded in 71% of patients, grade 3-4 thrombocytopenia in 10.5%, and grade 2 peripheral neuropathy in 2.6%. With a median follow-up period of 22 months for the patients who remain alive, 28 women are alive without recurrence and five are alive with persistent/recurrent disease. Five patients have died of disease. In the salvage setting, 45 women with persistent-recurrent disease after primary treatment were treated; 31 of these women had received prior radiation. In the salvage setting we observed 15 clinical complete responses, 15 partial responses, nine stable diseases, and six disease progressions. The objective response rate was 66.6%. Ten complete responders underwent subsequent surgery and seven had pathologic complete response (two in radiated areas). The response rate was 52% in radiated areas and 75% in nonradiated areas. The median survival time is 6 months for the nonresponders, 9+ month for the partial responders, and 13+ months for the complete responders. The most relevant side effect was myelotoxicity, with 91% of patients experiencing grade 3-4 myelotoxicity. One woman had life-threatening toxicity. This regimen yields a high response rate with acceptable toxicity and should be prospectively compared with other regimens. The high rate of pathologic complete and optimal responses might impact positively on survival, but only a longer follow-up period will allow objective assessment of this impact. The specific roles of paclitaxel and ifosfamide in this regimen remain to be fully understood.