OBJECTIVES Anti-TNF-alpha agents for Crohn's disease (CD) have good clinical efficacy but high acquisition cost compared to rival drugs. Previous modelling estimates are limited by a lack of primary data for costs of care and health state utility. The aim of the present study was to undertake an independent analysis incorporating data from recent trials and observational studies from the perspective of the UK National Health Service. METHODS Lifetime Markov analyses constructed to simulate outcomes and costs. CD was represented by 5 disease states: Full Response (CDAI<150, remission); Partial Response (CDAI 150-220, mild-to-moderate activity); Non-Response (CDAI>220, moderate-to-severe activity); Surgery and Death. The course of Crohn's disease under standard care was modelled from a transition matrix derived from the Olmsted county cohort. Systematic review identified ACCENT I (infliximab) and CHARM (adalimumab) as sources for efficacy data. We modelled an intention-to-treat strategy for biologics by including a non-responder cohort (representing patients excluded from the trials after failed induction). Surgical rates were based on observational data; cost estimates were taken from our UK dataset and utilities were derived for each state from an algorithm converting CDAI to EQ-5D utilities. RESULTS In the base-case analysis (lifetime horizon; 1-2 years continuous therapy; discount rate 3.5%) both agents achieved acceptable ICERs compared to standard care (Infliximab: £19,050 [1 yr]; £21,300 [2 yrs]; Adalimumab: £7,190 [1 yr]; £10,310 [2 yrs]). Lifetime therapy was dominated by standard care. Analyses over shorter time horizons, matched to treatment duration, resulted in unfavourable ICERs. CONCLUSIONS Contrary to earlier analyses, the model suggests acceptable ICERs for biological agents when considering a lifetime horizon with periods of at least 4 years continuous therapy. Apparent differences between rival biological agents must be interpreted cautiously as head-to-head trial data are not available.