OBJECTIVES: To track and categorise the use of decision modelling in the NICE appraisal process. Does it meet the needs of the Appraisal Committee and is it an efficient use of time and resources by all stakeholders? METHODS: A review was undertaken of the appraisals completed by NICE up to the end of October 2003. All available documentation from NICE (published and on the NICE website) was accessed and previous published commentaries were reviewed. Information was extracted from the use of decision models by: 1) manufacturers and sponsors in their submissions; 2) by the independent review groups assessing submissions; and 3) by NICE staff and the Appraisal Committee. The final NICE guidance to the NHS was reviewed to identify the importance of the model results in reaching a decision. RESULTS: Of 71 appraisals reviewed, only 5 (7%) involved no modelling. Manufacturers used models in their submissions in 87% of appraisals, and the independent review group developed its own additional model in 46%. In 30 appraisals (42%) both the review group and the manufacturers developed separate models. This was more likely to happen in recent appraisals. In the majority of cases, the review groups were critical of manufacturers’ models and in several appraisals the Appraisal Committee did not use the results of the review group’s own model. CONCLUSIONS: The timing of NICE appraisals makes modelling essential if the full range of costs and outcomes of a technology are to be considered. In appraisals of multiple products five different models may be produced. The potential for confusion and duplication of effort in the current process was highlighted by the recent review of NICE by WHO. The explicit and implicit cost of this duplication is considerable and alternative approaches to modelling should be considered.