Service-Oriented Architecture (SOA) provides a considerably new of invoking functionality. Service providers are to develop and deploy reusable services, and service consumers are to discover right services. A main difficulty in developing generic services is to design services generic enough so that the services can be effectively adapted to diverse service consumers. Moreover, this adaptation takes place at runtime, and hence dynamic adaptation and compositions are essential in service-oriented computing (SOC). For this, service variability among different consumers and contexts should be analyzed and designed into service components. In this dissertation, a service-oriented analysis and design (SOAD) process is presented, which focuses on modeling service variability and mismatch, engineering adaptability into service components and adapters, and enabling dynamic composition of services.