OBJECTIVES: There is a lack of understanding and empirical basis regarding what factors are important in the Medicare National Coverage Determination (NCD) decision making process. The objective of this study was to determine what factors predict favorable coverage decisions. METHODS: NCDs from 1999 through 2007 were reviewed using publicly available decision memoranda posted on the CMS website (n=140).  Data abstracted from decision memoranda were supplemented with cost-effectiveness information identified from an independent literature review.  When a decision memo included coverage decisions for multiple technologies or indications, an entry was made for each coverage decision in the memorandum.  The United States Preventative Services Task Force (USPSTF) guidelines were used to grade the supporting clinical evidence.  We created a dataset with the following variables: quality of supporting clinical evidence; availability of alternative interventions; cost-effectiveness of intervention; intervention type, and coverage requestor.  Logistic regression was used to determine what variables predicted favorable coverage. RESULTS: Good quality supporting clinical evidence was associated with an odds ratio (OR) of favorable coverage (95% CI) of 12.74 (3.02 – 53.74).  Interventions estimated to be dominant, i.e. less costly and more effective, or have an estimate of cost-effectiveness of