A total of 361 patients have been entered into four phase II trials in which gemcitabine was given as a 30 min infusion in a schedule weekly x 3 q4W, at starting doses of 800-1250 mg/m2. Three of these trials produced response rates of 22.5%, 20% and 21.8% (all response rates were independently evaluated by an extramural Oncology Review Board). One small study of 30 evaluable patients produced a disappointing low response rate of 3%, but in this study the mean and median dose delivered was under 700 mg/m2. Gemcitabine was well tolerated with modest levels of traditional cytotoxicities such as myelosuppression, nausea and vomiting, and alopecia. Pooled data from two of the studies show improvement in a number of disease-related symptoms. Objective response rates by prognostic factor were determined: stage IIIa (30.5%), stage IIIb (18.8%), stage IV (19.5%); performance status 0 (31.8%), 1 (16.7%) and 2 (21.7%); female gender (23.5%), male (17.5%); age < 70 (18.6%), age > 70 (21.9%). Gemcitabine can be considered for use as a single agent in patients unable or unwilling to tolerate combination chemotherapy. Dose-response data suggest that a dose of 1000 mg/m2 or more is required for optimal therapeutic effect. The single-agent activity of gemcitabine together with its non-overlapping toxicity and novel mode of action suggest that this agent should also be investigated in combination with other active agents in non-small cell lung cancer.