A massive personalized functional requirement oriented service composition method is presented to support service mass customization. In the realm of real-world service domains, there are usually massive customers who concurrently raise their personalized requirements. For such kind of scenarios, traditional approaches have to deal with each requirement one by one, consequentially leading to high cost. The proposed method considers massive requirements together. First, massive requirements are sorted based on the potential benefit. Then an iterative enhancement strategy is adopted to gradually enhance an existing composition solution based on the similarities between requirements. And finally, a service solution (called service network) with high customization degree is formed. Requirement consolidation is adopted to reduce the algorithm complexity. Comparisons made between the new method and two traditional composition approaches by experiments show the superiority of the new method. The method uses service network as a persistent infrastructure being a combination of personalization and standardization of services, and its objective is to select the minimal amount of candidate services to meet maximal amount of functional requirements, so as to achieve the best cost-effectiveness and high degree of customer satisfaction. The method can be applied to various modern service industries.