Modeling in animals is an invaluable tool in exploring the pathophysiology of human diseases and developing better therapies. Models can be generated using a variety of pharmacological, behavioral, and genetic approaches, but they all require extensive subsequent validation. Ideally, validation should be based on the following 3 axes: face validity (commonalties between the behavioral features of the model and of the human disorder being modeled), predictive validity (the specificity and degree to which drugs that are effective in humans have a corresponding effect in the model), and construct validity (a possible common mechanistic theory that can explain both the model and the human disorder being modeled). Most existing models for psychiatric disorders were developed from a face validity starting point, wherein a researcher noticed the appearance of a rodent behavior that was similar to a human pathological behavior and subsequently undertook investigations of predictive and construct validity. Some representative models developed in this manner include the hyperactivity (spontaneous or pharmacologically induced), sensitization, and sleep-deprivation models. In this review, we critically appraise the existing animal models for bipolar disorder, emphasizing their strengths and limitations. Furthermore, we discuss the technological advances that have led to an increased awareness of the roles of signal transduction pathways and neurotrophic cascades in the pathophysiology and treatment of bipolar disorder. New construct validity-driven models focusing on signaling pathways include models based on perturbations of G proteins, phosphoinositide signaling, and mitogen-activated protein (MAP) kinase cascades. These new models hold much promise in delineating the underlying pathophysiology of bipolar disorder and for the development of novel, improved therapeutics. Psychopharmacology Bulletin.