OBJECTIVES: We compared healthcare resource utilization (HRU) and costs associated with add-on therapy of RSG versus STG to MET. METHODS: Type II diabetes mellitus patients, ≥18 years, initiating RSG or STG (first dispensing = index date) add-on therapy with MET were identified in the PharMetrics database (1999-2008).  Patients were continuously enrolled for ≥6 months pre-index (baseline) and 12 months post-index, had ≥1 dispensing for MET in the 6-month pre-index period, did not use insulin or sulfonylurea, and were treated with RSG+MET before 05/01/2007 or STG+MET for ≥6 months post-index.  All-cause and diabetes-related HRU and annual costs ($2008) were reported for the 12-month follow-up period.  Workloss costs were estimated by applying hourly wage from the Bureau of Labor Statistics to missed work hours (hospitalization=8 hours; outpatient/emergency room [ER] visit=4 hours).  Multivariate analyses were conducted, adjusting for baseline demographics, comorbidities, and costs. RESULTS: Compared to STG+MET cohort (N=1,660) at baseline, RSG+MET cohort (N=3,731) was younger (55 vs. 58 years) with fewer comorbidities (Charlson Comorbidity Index [0.26 vs. 0.34]), and had lower total costs (RSG+MET: $7,875; STG+MET: $9,412; cost difference=$1,536, p=0.0043).  Frequency and costs of hospitalizations and ER visits were not different at baseline.  In the 12 months post-index period, all-cause HRU and corresponding annual cost difference between RSG+MET and STG+MET cohorts was enlarged (RSG+MET: $8,443; STG+MET: $10,757; cost difference=$2,314, p<0.0001).  After adjusting for covariates, the cost decrease associated with RSG+MET remained statistically significant (cost difference=$1,248; cost ratio=0.87; P=0.0120).  Diabetes-related adjusted incremental cost saving of RSG+MET over STG+MET was $599 (cost ratio=0.83, P=0.0160).  The adjusted workloss cost was also lower for RSG+MET compared to STG+MET (cost difference=$22, cost ratio=0.93; P=0.0120). CONCLUSIONS: Compared to the new DPP-4 agent STG combined with MET, RSG, a thiazolidinedione, combined with MET was associated with lower all-cause and diabetes-related direct healthcare costs and indirect workloss costs.