OBJECTIVES: 1) To identify key features of cost-effectiveness analyses (CEAs) of type 2 diabetes mellitus (T2DM) of the United States (US) population; 2) to assess the quality of T2DM CEAs; and 3) to identify the predictors of quality.  METHODS: We searched PubMed for several MeSH terms with English language restriction, through August 2011. The quality of eligible studies was evaluated using the Quality of Health Economic Evaluation (QHES) instrument.  Multiple linear regression analysis was conducted for the predictors of quality (overall QHES score) and independent variables being features of the T2DM CEA literature.  RESULTS: A total of 38 full-text articles met inclusion criteria of which:  Forty-six percent were pharmaceutical companies funded/sponsored, 82% were conducted from healthcare payers perspective, 77% were published in clinical-focused journals, 85% used quality-adjusted life-years (QALYs), 79% used published literature as the data source, 28% used the Center for Outcomes Research diabetes model, 51% were classified as disease treatment/management, and 64% used more than one-way sensitivity analysis. Overall, mean quality score using QHES was 73.2±11.5 and only 51% of studies scored≥75 (high-quality). Many studies (69%) failed to describe the analysis perspective and/or reasons for its selection; whereas, most of the studies (95%) used valid and reliable health outcomes scales/measures. Multiple linear regression found the following significant variables (p<0.05): journal impact factor (β=11.2, CI=7.4-14.0), studies using QALY (β=34.9, CI=11.2-48.1), and published after year 2000 (β=35.8, CI=13.9-48.6).  CONCLUSIONS: All studies funded/sponsored by a pharmaceutical company concluded the product of that company to be cost-effective; this may be indicative of publication bias and/or design bias. Several studies failed to follow the societal perspective recommendations of the US Panel on Cost-effectiveness in Health and Medicine, possibly because of preferences of the funding agency or researcher’s interests. Decisions based on these studies should consider quality and other key features of the later.