Objective: To profile physician prescription Behavior for patients with type 2 diabetes in a managed care setting. Methods: We used a retrospective cohort study design with patient-de-identified national health claims databases (70+ million unique lives from January 1, 2098 through December 31, 2004). Patients with T2D were identified based on ICD-9-CM codes, and were grouped into 6 cohorts (general practitioner, other primary care specialist, internist, endocrinologist, cardiologist, and other specialties) based on the practice specialty of physicians who prescribed the first insulin for their patients. For each cohort, we calculated proportions of patients with T2D using 7 individual insulin regimens (basal only, bolus only, basal-bolus, basal only with oral anti-diabetic medication (OAD), bolus only with OAD, basal-bolus with OAD, other). Chi-square test and Fisher's exact test were used to examine the association between patients' insulin regimens and the specialty of the physician who initiated their insulin use. Further we used canonical correspondence analysis to explore the association between practice specialty and patients' insulin regimens. Results: Both the Chi-square statistics (p<0.0001) and the Fisher's exact test (p<0.000001) indicate that patients' insulin regimens are strongly associated with physicians' practice type. The canonical correspondence analysis suggest that 86.27% of principal inertia could be explained by a dimension on which OAD use is correspond closely with general practitioners, endocrinologist, and internist practice specialties. The canonical correspondence analysis results also indicate that physicians' choice of insulin initiation regimen was statistically significantly associate with their practice specialties. Conclusion: The practice specialty of the physicians initiating insulin is strongly associated with their patients' overall insulin regimens. Compared to other physician practice specialties, endocrinologists were most likely to prescribe basal-bolus insulin with OAD, and least likely to prescribe basal only insulin.