Service composition is an emerging computing paradigm for constructing next-generation large-scale distributed applications within and across organizational boundaries. How to cope with the complexity of service composition resulting from increasingly sophisticated and heterogeneous services poses one of the great challenges for enterprises and organizations within many domains. This paper introduces a simple and novel service composition model based on functional programming and REST. We propose the λ @ -calculus which extends the classical λ-calculus to define identifier representing different resources and model a service as a function that consumes some identifiers and produces an result. A Functional Service Composition Language (FSCL) is also implemented based on our functional service composition model. The simplicity and optimization superiority of the proposed approach is verified by experimental results from a specific process in comparison with WS-BPEL.