BACKGROUND: This randomized phase II study compared the efficacy and toxicity between 4-week and 3-week schedules of gemcitabine monotherapy in advanced pancreatic cancer. METHODS: Patients with advanced pancreatic cancer were randomly assigned to either a 4-week schedule (gemcitabine at 1000 mg/m² as a 30-min infusion weekly for 3 consecutive weeks every 4 weeks) or a 3-week schedule (gemcitabine at 1000 mg/m² as a 30-min infusion weekly for 2 consecutive weeks every 3 weeks). The primary endpoint was the compliance rate during the first 8 weeks between the two groups. RESULTS: A total of 90 patients were enrolled. The compliance rate during the first 8 weeks was the same (53.3%). For the 4- and 3-week schedules, the tumor response rates were 14.2 and 17.1% (p = 0.92), median progression free survival was 112 and 114 days (p = 0.82), and median overall survival was 206 and 250 days (p = 0.84), respectively. Grade 3-4 neutropenia was the major adverse event in both schedules: 37.7 and 35.5% (p = 0.82). In contrast, thrombocytopenia (platelet count <70000/mm³) was significantly higher for the 4-week schedule: 26.6 and 4.4% (p = 0.008). The mean received dose intensity was equal: 588 and 550 mg/m²/week (p = 0.14). CONCLUSIONS: The 3-week schedule of gemcitabine did not improve the compliance rate during 8 weeks compared with the 4-week schedule, but it attained a comparable efficacy with lower toxicity. Further investigation will be needed to introduce it into daily practice. CLINICAL TRIAL REGISTRATION NUMBER: UMIN ID 974.