OBJECTIVES: NICE has been issuing health technology guidance based on cost-effectiveness evidence alongside other factors since 2000.  Previous studies have shown that technologies with higher incremental cost-effectiveness ratio (ICER) were more likely to be rejected; however none drew direct inference on the ICER threshold(s). Our aim is to directly estimate the ICER threshold(s) as well as the possible range. METHODS: Data were abstracted from the technology appraisals (TA) published from 03/2000 to 12/2012. For each decision to ‘recommend’ or ‘reject’ a technology we collected: ICER, publish date, disease area, technology type, comparator, reason for rejection if rejected, population and end of life (EOL) criteria (applies only to TAs after January 2009). Cancer related technologies which had been evaluated for EOL criteria were classified as satisfying or not satisfying the criteria. A Bayesian hierarchical model was implemented to estimate the overall threshold as well as the thresholds in different technology categories. RESULTS: 270 TA’s were evaluated.  After excluding those updated or terminated, a total of 187 appraisals with 323 decisions entered the final analysis. Non-informative priors were given to all the model parameters. The unadjusted estimate of the ICER threshold was £46,850 (95% CI: £40,420-£55,570). After adjusting for disease area, cancer related technologies had an estimated threshold of £48,550, (95% CI: £36,550-£63,200) compared to non-cancer related technologies’ estimated threshold of £43,430, (95% CI: £35,440-£52,300). Among the 37 technologies evaluated for end of life criteria, the estimated ICER threshold was £56,160, (95% CI: £39,020-£79,970) and £33,100, (95% CI: £19,180-£49,620) for those satisfying and not satisfying the criteria respectively. CONCLUSIONS: Preliminary assessment of NICE appraisals and associated ICER indicates that a likely ICER threshold exceeds the £20K-£30K quoted in the NICE Methods Guide.  Additional analyses are needed to assess the impact of other factors on the likely variability of ICER thresholds.