Fuel design is an important aspect that affects fuel reliability. For many years, in order to meet the unremitting pursuit of improving fuel reliability and the fuel management needs for long-term and high Burn-up, nuclear fuel vendors have optimized and improved their fuel design continuously. Fuel rod design involves many parameters which can make an influence on fuel rod performance under normal and transient conditions, and fuel rod performance must be met different criteria, such as temperature, internal pressure and strain criteria and so on, these criteria have no direct relationship, and the optimization parameters involved are restricting or even conflicting with each other. Applying traditional methods to improve fuel rod design is very difficult to optimize multiple parameters at the same time when there is only a range of design parameters given, and to meet the requirements of fuel rod design criteria. In this paper, a genetic algorithm is used to optimize fuel rod with many different design parameters, the design process could achieve fast convergence and we can obtain excellent solution set (fuel rod design parameters) with good distribution and diversity.