Testing workflow applications is important given their critical role in an organization's everyday operations. At the same time, systematic workflow application testing requires a significant amount of effort since these applications usually manipulate non-trivial data structures and interact with many other application components. Therefore, devising efficient and effective ways of generating test cases for workflow applications is a highly relevant problem. In this paper, we propose a method for workflow test case generation based on the Choice Relation Framework (CRF). The proposed method aims at reducing the amount of input required from the sofware tester, while allowing test cases to be generated incrementally. To this end, test cases are generated in such a way as to cover the most frequently-occurring workflow execution traces, which are identified based on conditional branching probabilities typically available in workflow models. An empirical evaluation demonstrates that the proposed method significantly reduces the amount of input that testers need to provide in order to produce test cases that cover the most frequent paths of the process.