OBJECTIVES: A number of studies have reported that patients with type 2 diabetes have higher annual mean expenditures than those without this condition.  Additionally, the range of costs among these patients is quite large. Studies have also illustrated that certain acute events and co-morbidities drive excess mean costs. Whether these events and conditions affect cost in the same way across the distribution has not been studied.  In order to more fully understand potential drivers of costs, we performed a descriptive analysis across quartiles of the cost distribution.  METHODS: Data for this study come from a US health plan affiliated with i3 Innovus. We included members aged 18 or older that had evidence of type 2 diabetes over the period January 1, 2004 to December 31, 2006. An index date was defined as the date of the earliest qualifying medical or pharmacy claim. Patients were required to have continuous enrollment two years prior to (baseline period) and 2 years following (follow-up period) the index date. Variables were created for the following categories: demographics characteristics, diagnoses, medications, procedures and clinical markers (e.g. lab values for HbA1C/Lipids).  Differences in potential cost drivers across the quartiles of the cost distribution were assessed. RESULTS: Mean annual cost for those in the highest quartile was 6X higher compared to those in the lowest quartile ($3,200 versus $19,700).  Although there did not appear to be differences in HbA1c and lipid levels across the quartiles, meaningful differences were seen in many of the other variables analyzed.  For example, “diseases of the heart” ranged from 17% in the lowest quartile to 42% in the highest quartile. CONCLUSIONS: The study illustrates that a quantile-based analytical approach may allow for a deeper understanding of the drivers of health care costs.