OBJECTIVES: To assess the clinical effectiveness and cost-effectiveness of gemcitabine, used in combination with paclitaxel, as a second-line treatment for people with metastatic breast cancer who have relapsed following treatment with anthracycline-based chemotherapy. DATA SOURCES: Electronic databases were searched from inception to March 2006. Clinical advisers were also consulted. REVIEW METHODS: A systematic review of the literature was undertaken to appraise the clinical and cost-effectiveness of gemcitabine. A Markov state transition model was developed for the economic evaluation. RESULTS: The systematic review identified only one randomised controlled trials (RCT), and this has not yet been fully published. The methodological quality and quality of reporting of the included trial were assessed to be poor using standard criteria, but this may be due to the lack of information in the limited publications rather than being a fair reflection of the trial's quality. This RCT compared gemcitabine and paclitaxel therapy with paclitaxel monotherapy in 529 patients with metastatic breast cancer who had previously received anthracyclines, but no prior chemotherapy for metastatic breast cancer. Approximately 71% of the gemcitabine/paclitaxel patients survived for 1 year, compared with 61% of the paclitaxel group. The hazard ratio showed a 26% lower chance of survival in the paclitaxel group, and time to progressive disease was also shorter in this group. The overall response rate was higher in the gemcitabine/paclitaxel group than in the paclitaxel group. Adverse events, particularly neutropenia, were more common with gemcitabine/paclitaxel combination therapy than with paclitaxel therapy alone. The economic model was run for a simulation of 1000 patients, assuming that chemotherapy continued until patients' disease progressed. This base-case analysis found an incremental cost-effectiveness ratio (ICER) of 58,876 pounds per quality-adjusted life-year (QALY) gained and 30,117 pounds per life-year gained. The model was re-run with treatment restricted to a maximum of six cycles per patient, reflecting normal practice. This yielded an ICER of 38,699 pounds per QALY gained and 20,021 pounds per life-year gained. CONCLUSIONS: The review of clinical effectiveness is based on data from a single RCT that has not yet been fully published. While only tentative conclusions can be drawn from this, the evidence may indicate that treatment with gemcitabine and paclitaxel confers an improved outcome for patients in terms of survival and disease progression, but at the cost of increased toxicity. An economic model developed for this review reflects high costs per QALY for this treatment combination. The base-case analysis shows high ICERs, with costs per QALY gained close to 60,000 pounds. Adopting a more realistic treatment protocol, with chemotherapy limited to a maximum of six cycles, gives a more favourable cost-effectiveness estimate. However, this was still higher than would usually be considered to be a cost-effective treatment from the NHS's perspective. Future research recommendations include an update of this review in 12-18 months' time, by which time the included RCT should be fully published. It would also be useful to compare gemcitabine with currently used treatments for metastatic breast cancer, including capecitabine and vinorelbine.