Few clinical coaching studies are both endorsed by real cases and focused on reducing suboptimal diabetes control. We evaluated the effectiveness of coaching on improving type 2 diabetes goals after 3 years of implementation in primary care. A cross-sectional study with follow up was conducted during 2008-2011. Coaching consisted of guiding family doctors to improve their clinical abilities, and it was conducted by a medical doctor trained in skill building, experiential learning, and goal setting. Effectiveness was assessed by means of fasting plasma glucose and glycosylated hemoglobin outcomes. The main analysis consisted of 1×3 and 2×3 repeated measures ANOVAs. A significant coaching×time interaction was observed, indicating that the difference in glucose between primary care units with and without coaching increased over time (Wilks' lambda multivariate test, P