In the future vision of software engineering, services from world-wide markets are composed automated in order to build custom-made systems. Supporting such scenarios requires an adequate service matching approach. Many existing approaches do not fulfill two key requirements of emerging concepts like On-The-Fly-Computing, namely (1) comprehensiveness, i.e., the consideration of different service views that cover not only functional properties, but also nonfunctional properties and (2) fuzzy matching, i.e., the ability to deliver gradual results in order to cope with a certain extent of uncertainty, incompleteness, and tolerance ranges. In this paper, I present a fuzzy matching process that distinguishes between different fuzziness sources and leverages fuzziness in different matching steps which consider several service views, e.g., behavior and quality properties.