BACKGROUND: Severe depression can increase the risk of psychiatric hospitalization, as well as inpatient and outpatient care; it may also lead to long-term absenteeism from work. However, the cost-effectiveness of antidepressant therapy for severe depression has been little studied. OBJECTIVE: The aim of this work was to investigate the cost-effectiveness of escitalopram compared with citalopram in patients with severe depression (Montgomery-Asberg Depression Rating Scale [MADRS] total score > or = 30) in the United Kingdom. METHODS: A probabilistic decision tree with a 6-month time horizon was adapted to the UK setting. The model incorporated clinical data, resource use directly related with care of severe depression, and lost productivity costs due to absenteeism. Primary results were remission (MADRS < or = 12) and costs (in year-2003 British pounds [1.00 British pound = 0.62 US dollars in January 2003]) of treatment calculated from the perspectives of UK society and the National Health Service (NHS). Secondary outcome was first-line success of treatment (ie, remission [MADRS < or = 12] without switch of drug). Remission, discontinuation, and response rates were derived from a meta-analysis of 506 patients with severe depression and extrapolated to 6 months. Standard UK price lists and literature were used to identify costs of resources. Societal costs of lost productivity were calculated using the human capital approach. RESULTS: Treatment of patients with escitalopram instead of citalopram rendered a higher overall remission rate (relative difference, 10.3%) and first-line success rate (relative difference, 35.4%). The mean cost per successfully treated patient was 15.7% (146 British pounds) lower for escitalopram (786 British pounds [range, 702-876 British pounds]) compared with citalopram (932 British pounds [range, 843-1028 British pounds]) from the NHS perspective and 15.6% (238 British pounds) lower for escitalopram (1283 British pounds [range, 1157-1419 British pounds]) than for citalopram (1521 British pounds [range, 1383-1675 British pounds]) from the societal perspective. The mean cost per severely depressed patient treated (overall study group) was 32 British pounds lower for escitalopram (422 British pounds [range, 404-441 British pounds]) than citalopram (454 British pounds [range, 436-471 British pounds]) from an NHS perspective and 50 British pounds lower for escitalopram (690 British pounds [range, 665-714 British pounds]) than citalopram (740 British pounds [range, 715-767 British pounds]) from the societal perspective. Using multivariate sensitivity analyses, we found that, in 99.8% of the cases, escitalopram was dominant from both perspectives at all ranges of probabilities tested. A sensitivity analysis on the acquisition cost of citalopram verified that, from the societal perspective, escitalopram remained the dominant strategy, even at a cost of 0.00 British pounds for citalopram. CONCLUSIONS: These results suggest that escitalopram is a cost-saving alternative to citalopram for the treatment of severe depression in the United Kingdom from the perspectives of both the NHS and society. Therefore, a possible advantage may exist at the population level in the treatment of severe depression with escitalopram in the United Kingdom.