This paper presents a way to represent and solve the problem of graph matching – also known as the subgraph homomorphism problem – as a constraint satisfaction problem (CSP), opening up direct access to the large variety of research findings on optimized solution algorithms for CSPs. By decoupling the solution algorithm from the concrete graph model, this approach allows for variations of the model without affecting the algorithm. Furthermore, complementing the standard CSP definition, a query concept is introduced to allow for abstract representation of concrete implementation properties for optimization purposes.