Evaluation of railway transportation efficiency is one of the most important issues of railway transportation industry. National Iranian Railway includes 15 operational regions which operate under supervision of Iranian Railway headquarters. In this paper, we have applied Data Envelopment Analysis (DEA) to measure the efficiency of these 15 railway regions (DMUs) both in Variable and Constant Returns to Scale. To achieve this goal we have applied two models with different outputs. The second model contains railway accidents and fatal data. We have also ranked the efficient railway regions by applying a heuristic ranking technique to find out the most efficient one. Consideration of desired and undesired outputs is a new approach in efficiency evaluation of railway regions that has more reliable results. The results reflect the necessity of considering the undesired outputs for a more realistic overall efficiency evaluation and accordingly, appropriate corrective recommendations are presented for improving the performance of inefficient railway regions.