A randomized phase III study was conducted to assess the addition of molgramostim (GM-CSF) to the combination of methotrexate, vinblastine, doxorubicin, and cisplatin (MVAC) in terms of response rate, progression-free survival, and survival in women with advanced, recurrent, or metastatic carcinoma of the cervix or vagina. Patients received four 4-week cycles of methotrexate 30 mg/m2 IV days 1, 15, 22; vinblastine 3 mg/m2 IV days 2, 15, 22; doxorubicin 30 mg/m2 IV day 2; and cisplatin 70 mg/m2 IV day 2 with or without GM-CSF 5 microg/kg every 12 hours subcutaneously days 3 to 12. They were then reevaluated for operability. Those who were not surgical candidates were offered additional chemotherapy until progression or toxicity. Those who were surgical candidates were offered surgical resection of remaining tumor followed by involved-field external beam irradiation to sites of no prior irradiation and intraoperative irradiation to sites of prior external beam irradiation. This trial closed after 36 eligible patients were entered because of poor accrual. Although more than 40% of patients on each arm received fewer than four cycles of MVAC, the clinical response rate was 78% (95% CI: 52-94%) and 50% (95% CI: 26-74%) for MVAC and MVAC + GM-CSF, respectively; the median time to progression was 10.2 and 11.8 months, respectively; and median survival was 13.8 and 16.0 months, respectively. Toxicity was substantial, with more than 40% experiencing grade III to IV leukopenia, and nearly 40% experiencing grade III to IV stomatitis. MVAC with or without GM-CSF support achieves high response rates in patients with advanced, recurrent, or metastatic cervical carcinoma despite dose reductions and deletions. Its progression-free survival and overall survival rates appear promising. These results need to be confirmed within a large randomized phase III clinical trial.