OBJECTIVES: The present study examined the association between initial monotherapy with metformin (MET) or sulfonylurea (SU) and subsequent initiation of insulin in older patients with type 2 diabetes mellitus (T2DM). METHODS: In a retrospective cohort study using the GE electronic medical database, eligible patients with T2DM included those ≥65 yrs who received their first prescription (Rx) of SU or MET as initial monotherapy between 2003 and 2008. The follow up lasted to the end of 2009 or the patient's latest data available. Insulin initiation was determined by Rx records. Logistic regression analysis evaluated the likelihood of insulin addition and Cox regression model estimated time to initiation of insulin. Differences in baseline characteristics (demographics, clinical and lab measures, and comorbidities) were controlled for using propensity score matching (PSM). RESULTS: Overall, 12,036 patients were included in the analysis with 6,018/group. Mean age was 75 years and 50% were male. While controlling for differences in baseline characteristics using PSM, patients who initiated with SU had a significantly (p<0.001) higher incidence of insulin addition (2.8% vs. 1.4%) compared to those initiated with MET after 1 year of follow up. The likelihood of add-on insulin use was higher in patients initiated with SU than with MET (OR [95% CI] = 1.96 [1.51, 2.55]; p<0.001). SU was also significantly associated with shorter time to insulin add-on use compared to MET (HR [95% CI] = 2.20 [1.91, 2.52]; p<0.001). The rate of insulin addition remained significantly higher (p<0.001) with initial SU monotherapy vs. MET after 2 and 3 yrs of follow up (6.1% vs. 2.6%, and 8.1% vs. 3.9%, respectively). CONCLUSIONS: In a cohort of older patients with T2DM initiating antihyperglycemic therapy, patients who started with SU monotherapy received insulin therapy more frequently and earlier than those who started with MET.