Model transformation is the key technique of model driven development. To handle difficult problems, it is necessary to combine smaller transformations into a complex one. Due to the heterogeneity among different transformation techniques, it is difficult to combine them together. The paper analyzes four essential conditions for composite transformations. Next, the paper proposes a composite transformation model, which consists of common type representation, common model representation, common transformation description, and composite transformation definition, in order to realize composite transformations. The paper also introduces the design and implementation of a transformation composition framework. A case study is also presented to illustrate the feasibility of this approach.