Data Envelopment Analysis (DEA) approach was used to estimate technical efficiency and followed by regressing the technical efficiency scores to farm specific characters under tobit regression model. Primary data was collected from random samples of 240 (120 from each) coffee famers. Mean technical efficiency score was 0.89 and 0.83 in organic and conventional coffee farming respectively. Farms operating under constant return to scale (CRS), decreasing return to scale (DRS) and increasing return to scale (IRS) were 31.67, 3.83 and 37.5% respectively in organic coffee and 29.17, 25 and 45.83% respectively in conventional farming areas. These scale technology defines a production set that is closed and convex with property of strong disposability. Tobit regression showed the variation in technical efficiency was related education, farm experience and training/extension services and access to credit. Farmers would reconsider the rationing of input and learn from technically efficient farms practices. Policy implication will rest on production planning strategy.