The efficacy and safety of gemcitabine were investigated in 16 patients with unresectable pancreatic cancer (Arm A), compared with 16 patients who received chemotherapy without gemcitabine (Arm B) and 44 patients who received best supportive care (Arm C). A gemcitabine 30 min i.v. infusion at a starting dose of 1,000 mg/m2 was administered once a week for 3 weeks with a 1 week rest. Dose reduction and cycle delay were applied because of toxicity in 62.5% of the cases in the first cycle and 31.3% in the second cycle. Hematological toxicity was observed in 81.3%, nausea/vomiting in 37.5% and fatigue in 18.8%. Clinical benefit response was observed in 25.0% in Arm A, as compared with the lower rate of 6.25% in Arm B. Response rates were comparable. The median time of outpatient treatment was 98.5 days in Arm A and 34.0 days in Arm B, respectively. The median survival time was 200 days in Arm A, 121 days in Arm B and 82.5 days in Arm C, respectively. In Arm A, the higher dose intensity showed a longer survival time. The results show that gemcitabine can be administered in outpatient clinics with dose reduction and cycle delay, and that higher dose intensity generates clinical benefit, survival advantage and prolonged outpatient treatment time.