With the excessive growth of complex information systems operating in dynamic environments, the need for managerial policies to conduct such systems is an essence. Moreover software engineering challenges are moving towards design and implementation of robust software systems, self adaptive software systems and in general systems that can adapt themselves to changing customer’s requirements and stochastic environmental variables. Service Oriented Architecture (SOA) is introduced to address these issues. For this means, SOA has brought about new concepts such as services, dynamic binding, loose coupling, and well defined interfaces together to present a new paradigm in which adaptation and quality of services (QoS) are highlighted. SOA must provide adaptation techniques in order to guarantee quality of service in case of services being unavailable, unreachable or working with a lower rank in comparison with their acclaimed QoS. To be able to do so, SOA must be enriched by a feature called self adaptation. Self adaptive software systems are those who can adapt themselves with the changes in system environments and costumer’s needs. These systems follow a strategy first designed by IBM Corporation which uses a feedback along with a controller element. The controller element consists of a monitor to detect symptoms, an analyzer to analyze what has been monitored, a planner to plan for changes based on analysis and an executer to effect the changes through system. While many recent researches focused on presenting adaptive SOA which operates in run time, this thesis tries to answer the main question: “Ithereaframework(oraetofthem) which supports the discovery, selection, adaption and monitoring of services based on the Quality of Service?” To reach the answer of this question, a framework named MQA-SOA (Multilevel QoS Aware Service Oriented Architecture) is presented which performs several adaptation methods to guarantee and control QoS in both run time and design time. MQA-SOA as its name suggests operates in two separate phases of SOA: modeling phase maintenance and deployment phase. In maintenance and deployment phase two conventional concepts; service consumer and service broker are refined in order to be able to send, receive and store feedbacks. Service consumers make use of these feedbacks to update their service trees – another extension of service consumers – which enables them to effectively select the services with optimum QoS. Service trees are constructed based on red and black tree structure due to its quick operations. In modeling phase, a novel model driven method for service identification is introduced which identifies business aligned services regarding f Keywords: Service Oriented Architecture, Self Adaptive Systems, Service Identification, Dynamic