The assembly line balancing problem has been focused by many research works because the efficient management of the assembly line might influence not only the quality of the products but also the working conditions for the workers. This paper deals with U-shape mixed-model assembly line balancing and considers both the processing time and the physical workloads. We suggest the goal programming approach for this situation and to overcome some difficulties of finding optimal solution, we adopt the genetic algorithm that is one of the most promising solution techniques. We tested several test problems and present the results that indicate some improvement for the line balancing as well as the stable performance of the algorithm.