Some animal models of depression, including the majority of the more recently introduced models, are better characterized as models of predisposition to depression. In the first part of this paper, we show that the basis for such a model could be either a procedure that increases the ease with which an analogue of major depression may be evoked, or a presentation analogous to dysthymia (chronic mild depression). We then consider how the concepts of predictive, face, and construct validity apply to such models. Next, we review the validity of the available models of predisposition to depression, which derive from genetics, genomics, developmental manipulations, and brain lesioning. Finally, we compare the performance of the different models, using a novel scoring system that formalizes the evaluation of animal models against each of the three sets of validation criteria.