As with other software development artifacts, model transformations are not bug-free and so must be systematically verified. Their nature, however, means that transformations require specialist verification techniques. This paper brings together current research on model transformation verification by classifying existing approaches along two dimensions. Firstly, we present a coarse-grained classification based on the technical details of the approach (e.g., testing, theorem proving, model checking). Secondly, we present a finer-grained classification which categorizes approaches according to criteria such as level of formality, transformation language, properties verified. The purpose of the survey is to bring together research in model transformation verification to act as a resource for the community. Furthermore, based on the survey, we identify a number of trends in current and past research on model transformation verification.