OBJECTIVES: CML is a malignant orphan disease of the blood and bone marrow. Imatinib 400mg (up to 800mg with inadequate responses) daily is currently recommended for treatment of newly diagnosed patients or after failure with interferon-α. Dasatinib 100 mg daily has been shown to offer significant clinical efficacy in patients failing imatinib. Its cost-effectiveness compared to imatinib 600/800 mg has not been assessed in this patient group. METHODS: A partitioned survival/costing model was developed to estimate the lifetime costs and benefits associated with dasatinib and imatinib from a UK health service perspective using a lifetime horizon and monthly cycles. Prognosis was assigned for dasatinib and imatinib patients to each of five initial best clinical response categories at 12 months. Response category specific survival was based on long-term data from IRIS clinical trial and response rates from a phase III randomized study. Utility and resource use data were taken from recent UK based studies and all unit/drug costs were taken from appropriate national databases and discounted at 3.5%per annum. Probabilistic and deterministic sensitivity analyses were conducted to estimate the confidence around the results. Outcomes are reported via incremental cost-effectiveness ratios (ICERs); benefit is expressed as quality adjusted life years (QALYs). RESULTS: Compared to imatinib, dasatinib offered an additional 3.53 QALYs but incurred £90,800 of additional costs. The ICER was therefore £25,700/QALY gained. At a threshold of £30,000/QALY gained, dasatinib had a 98.1% probability of being cost-effective. Deterministic analysis showed that the model was sensitive to changes in 12 month response probabilities and drug costs. The model was robust to changes in adverse event rates/ costs, and to utility estimates. CONCLUSIONS: Dasatinib has been shown to be clinically superior to imatinib in CML patients who have failed imatinib treatment and is a cost-effective alternative to imatinib dose escalation in this patient group.