The growth rate of poultry meat production is not sufficient to satisfy the per capita meat requirement in Bangladesh. Therefore ways to improve the efficiency of the existing production technologies must be investigated if these demands are to be met. This review examines ways this could be achieved. The technical, allocative and economic efficiency of poultry meat production based on farm level survey data was estimated using the Data Envelopment Analysis (DEA) approach. The study used 100 commercial poultry farms in Bangladesh for the calendar year of 2007. The results demonstrated that there is substantial technical, allocative and economic inefficiency in poultry production in Bangladesh. Under the constant return to scale (CRS) specification, technical, allocative and economic efficiencies were 88, 70 and 72% respectively, whereas under the variable returns to scale (VRS) specification those efficiencies were 89, 73, 66% respectively. Thus, the results indicate that efficiencies varied substantially across the sample farms. In order to attempt to explain some of these variations, the efficiency scores were regressed based on farm specific variables such as the farmer's age, education, experience, total landholdings, poultry farm size and level of training using a Tobit regression framework. The estimated DEA model identified that there is great potential for increasing poultry farm efficiency using the existing level of inputs and resources more efficiently. Specifically it showed that the level of education and training the farmer receives are two of the most important factors contributing to the variations in efficiency seen during this study. These research findings are valuable to policy makers and extension workers in order to guide policies towards increasing efficiency.