Model transformations are crucial for the success of Model Driven Engineering. Testing is a prevailing technique of verifying the correctness of model transformation programs. A major challenge in model transformation testing is the oracle problem, which refers to the difficulty or high cost in determining the correctness of the output models. Metamorphic Testing alleviates the oracle problem by making use of the relationships among the inputs and outputs of multiple executions of the target function. This paper investigates the effectiveness and feasibility of metamorphic testing in testing model transformation programs. Empirical results show that metamorphic testing is an effective testing method for model transformation programs.