Heavy oil and tar sands resources comprise about 70 percent of the world's oil reserves and this reservoirs can offset the declining production from conventional reservoirs. Thermal enhanced oil recovery (EOR) methods are employed to exploit the huge reserves of heavy oil due to their high viscosity values. Thermal processes aim to increase its mobility in order to improve its production. Among these methods, the steam-assisted gravity drainage (SAGD) is one of the most efficient techniques. In this method, two horizontal wells are drilled and hot steam is injected from a well to move oil toward the other well. Optimization of operating parameters during this process is very important. The injection rate or pressure control of wells are the most common EOR methods. In this paper for the first time, in addition to the injection rate of the injector and production wells, the steam injection temperature is also optimized. It was shown that there is an optimum amount for the temperature of injected steam. In addition entropy generation analysis was performed for different cases. To simulate the process, a commercial software was used and optimization of operating parameters is performed using the pattern search algorithm. Entropy generation calculated based on the results of numerical simulations using a computational code has been written for this case. The results show that the maximum oil production corresponds with the minimum entropy generation number and thus the entropy number can be used as an appropriate objective function in order to enhance oil recovery.