A number of path selection criteria have been proposed throughout the years. Unfortunately, little work has been done on comparing these criteria. To determine what would be an effective path selection criterion for revealing errors in programs, we have undertaken an evaluation of these criteria. This paper reports on the results of our evaluation of path selection criteria based on data flow relationships. We show how these criteria relate to each other, thereby demonstrating some of their strengths and weaknesses. In addition, we suggest minor changes to some criteria that improve their performance. We conclude with a discussion of the major limitations of these criteria and directions for future research.