Although there has been recent research into ways to design visual programming languages and environments, little attention has been given to systematic testing in these languages, and what work has been done does not address "power" features such as recursion. In this paper, we discuss two possible ways the "What You See Is What You Test" methodology could be extended to accomodate recursion. The approaches are presented in terms of their testing theoretic aspects and then implementation strategies and algorithms. Since the goal is to help the people using these languages, we also present an empirical study and use its results to inform our choice as to which of the two approaches to adopt.