OBJECTIVE: This analysis attempts to highlight the differences in response rates and toxicity between Gemcitabine combined with a platinum-based therapy and other combinations of platinum-based chemotherapy, in the treatment of non-small-cell lung cancer (NSCLC). METHODS: This pooled analysis used summary statistics from clinical trials published up to December 2001. The analysis pooled odds ratios (OR) and associated confidence intervals (CI) using a fixed-effects model. The efficacy outcomes considered are responses (both partial and complete) and progressive disease. Grade 3 and 4 toxicities are considered using the WHO criteria for the following adverse events: alopecia, nausea and vomiting, anaemia, neutropenia, thrombocytopenia and neuropathy. RESULTS: Patients receiving Gemcitabine combined with a platinum therapy are more likely to experience a response to treatment than patients receiving other platinum based combinations. The OR for complete and partial responses is 2.68, (CI 1.53-4.67) and is 0.44 for progressive disease (CI 0.32-0.59). Gemcitabine patients experienced fewer cases of alopecia (OR 0.15, CI 0.10-0.22) and neutropenia (OR 0.6, CI 0.47-0.77). In contrast, Gemcitabine patients experienced a greater number of grade 3 or 4 anaemia (OR 1.92, CI 1.41-2.61) and thrombocytopenia (OR 6.76 CI 4.95-9.23) incidences. For neuropathy and nausea and vomiting there was no evidence for any of the chemotherapies having fewer patients experiencing toxicities. CONCLUSIONS: The implications of this analysis at the patient level is that if response is of primary importance, then on a purely clinical basis Gemcitabine should be the treatment of choice. Gemzar based chemotherapy had a higher number of responses and fewer adverse events for alopecia and neutropenia. To validate these results, a meta-analysis should be conducted with stratification for key variables using patient level data.