A data envelopment analysis of a sample of 76 Nepalese rice farmers reveals average relative economic, allocative, technical, pure technical and scale inefficiencies as 34, 13, 24, 18 and 7 per cent, respectively. The significant variations in the level of inefficiency across sample farms are attributed to the variations in the ‘use intensities’ of resources such as seed, labour, fertilisers and mechanical power. In addition, a second stage Tobit regression shows the variation is also related to farm‐specific attributes such as the farmers’ level of risk attitude, the farm manager's gender, age, education and family labour endowment. Based on the empirical findings, policy implications and development strategies for improving efficiency of Nepalese rice farms are briefly discussed.