OBJECTIVES: Understanding current and new paths of treatment (POT) in diabetes is pivotal for estimating budget impacts when new antidiabetic agents are introduced to pharmacy formularies.  Although diabetes treatment guidelines exist to guide prescribers, very limited real-life data are available to illuminate actual POT’s for patients initiating insulin treatment. The aim of this study was to establish the most prevalent POT’s for insulin initiation using real-world US pharmacy data. METHODS:   Medicaid, Cash and 3rd party payer pharmacy claims were obtained from SDI. A patient’s first insulin Rx between October and December. 2004 served as an index Rx. A 6 month look-back identified prior treatments and a 3 year look-forward tracked changes from the index Rx. To be eligible for analysis, patients had to have no insulin Rx in the look-back and be in the SDI database in the first and last quarter of observation. The pattern of payers and prescribers were identified along with the most frequent Rx combinations for each step. The average duration for each step was analyzed. RESULTS: Out of 351,887 patients with an index insulin Rx, 78,213 had no insulin in the look-back. Claims were paid by Medicaid (13%), cash (4%), and 3rd party (83%). The top-10 combinations prior to the index insulin constituted 96% of the data. The top 3 were Metformin(MET)+Sulfonylurea(SU) (18%), Met alone (17%) and SU alone (13%). The top-10 combinations initiating index insulin constituted 59% of the data. The top 3 were MET+SU+basal insulin (11%), MET+SU+TZD+basal insulin (11%), and SU+basal insulin (6%). On average, patients initiating insulin went 230 days without changing their therapy. CONCLUSIONS: Based on real-world US data, a small number of therapy combinations explain the POT’s before insulin initiation. Upon initiating insulin treatment, combinations are more varied, although most include basal insulin.