Considering that there are three aspects of constrains in the service selection process, such as control structure within a composition plan, relationship between concrete services, and tradeoff among multiple QoS indexes, a QoS based optimal Web services selection method by multi-objective genetic algorithm is presented. First we design a chromosome coding method to represent a feasible service selection solution, and then develop genetic operators and strategies for maintaining diversity of population and avoiding getting trapped in local optima. Experimental results show that within a finite number of evolving generations this algorithm can generate a set of nondominated Pareto optimal sol.utions which satisfy to user's QoS requirements.