There currently exists no consensus as to the cause of the ice ages of the late Pleistocene. Many of the competing hypotheses have been formulated into mathematical models which enables a rigorous comparison to be made. Spectral analysis fails to distinguish rival models. Using regression analysis, we examine the relative performance of several models, each representative of a different type of modeling approach, as explanations of both the global ice volume and also the time rate of change of the global ice volume. We find there is no objective evidence in the record in favor of any particular model. The respective merits of the different theories must therefore be judged on physical grounds. ¿ 1999 American Geophysical Union