The decisions made by the National Institute for Clinical Excellence (NICE) give rise to two questions: how is cost-effectiveness evidence used to make judgements about the ‘value for money’ of health technologies? And how are factors other than cost-effectiveness taken into account? The aim of this paper is to explore NICE's cost-effectiveness threshold(s) and the tradeoffs between cost effectiveness and other factors apparent in its decisions. Binary choice analysis is used to reveal the preferences of NICE and to consider the consistency of its decisions. For each decision to accept or reject a technology, explanatory variables include: the cost per life year or per QALY gained; uncertainty regarding cost effectiveness; the net cost to the NHS; the burden of disease; the availability (or not) of alternative treatments; and specific factors indicated by NICE. Results support the broad notion of a threshold, where the probability of rejection increases as the cost per QALY increases. Cost effectiveness, together with uncertainty and the burden of disease, explain NICE decisions better than cost effectiveness alone. The results suggest a threshold somewhat higher than NICEs stated ‘range of acceptable cost effectiveness’ of £20 000–£30 000 per QALY – although the exact meaning of a ‘range’ in this context remains unclear. Copyright © 2004 John Wiley & Sons, Ltd.