This paper addresses the problem of generating symbolic test cases for testing the conformance of a black-box implementation with respect to a specification, in the context of reactive systems. The challenge we consider is the selection of test cases according to a test purpose, which is here a set of scenarii of interest that one wants to observe during test execution. Because of the interactions that occur between the test case and the implementation, test execution can be seen as a game involving two players, in which the test case attempts to satisfy the test purpose. Efficient solutions to this problem have been proposed in the context of finite-state models, based on the use of fixpoint computations. We extend them in the context of infinite-state symbolic models, by showing how approximate fixpoint computations can be used in a conservative way. The second contribution we provide is the formalization of a quality criterium for test cases, and a result relating the quality of a generated test case to the approximations used in the selection algorithm.