To examine sustained effects of an educational intervention, we repeated a successful quality improvement (QI) project on medication safety and cost-effectiveness. In October 2007 and August 2008, facility leadership and geriatrics faculty identified all patients receiving ≥9 medications (polypharmacy cohort) in a 170-bed teaching nursing home and taught Geriatric Medicine fellows (n=12 in 2007, 11 in 2008) to: 1) systematically collect medication data; 2) generate medication recommendations (stop, taper, or continue) based on expert criteria (Beers Criteria) or drug-drug interaction programs; 3) discuss recommendations with patients’ attending physicians; and 4) implement approved recommendations. Over the two projects, the polypharmacy cohorts demonstrated decreased potentially inappropriate medications (odds ratio (OR) 0.78, 95% confidence interval (95%CI)0.69–0.88, p<0.001), contraindicated medications (OR=0.63, 95%CI=0.47–0.85, p=0.002) and medication costs (OR=0.97, 95%CI=0.96–0.99, p<0.001). Our findings suggest that programs planning educational QI projects for trainees may benefit from a multi-year approach to maximize both clinical and educational benefits.