Circumplex representations of data have enjoyed widespread popularity in personality and social psychological research. In this article we review the conceptual assumptions implied by circumplex representations and we discuss the limitations of traditional statistical methods for testing these assumptions. A relatively nontechnical overview of a new covariance structure modeling approach to testing circumplex structure is provided. The use of this approach is illustrated with two published data sets. The advantages and disadvantages of this approach relative to more traditional statistical approaches are discussed. The conclusion is that the covariance structure modeling approach has significant advantages in that it provides a closer conceptual match to the theoretical assumptions of circumplex representations, supplies information more directly relevant to circumplex representations and permits more precise and flexible testing of hypotheses derived from circumplex representations.