This study measures efficiency of Aman rice farms and assesses the effect of farm-specific variables on the efficiencies using cross-section data in Bangladesh. Data envelopment analysis (DEA) method is used to measure the efficiency levels of Aman rice farms in Bangladesh. We used multi-stage sampling technique to collect the data of 455 rice farms using face to face interview method. The results show that the mean overall technical efficiency, pure technical efficiency, and scale efficiency are 0.75, 0.91, and 0.83, respectively. The beauty of this article is the use of scale efficiency which identifies the three different stages of production. The finding of the scale efficiency shows that about 0.89 percent of the Aman rice farm is operating at economies of scale that is really higher than optimal level. The Tobit regression model is employed in order to assess the cause of inefficiencies of rice farms. The results of Tobit regression implies that agriculture policy variable, farmers having bank account for receiving government agricultural assistance, positively relates to the technical efficiency of Aman rice farms. The policy implication of these findings implies government assistance such as cash incentive, and extension service need to be extended and implemented with proper targeting.