This paper evaluates the validity, reliability and utility of the chronic mild stress (CMS) model of depression. In the CMS model, rats or mice are exposed sequentially, over a period of weeks, to a variety of mild stressors, and the measure most commonly used to track the effects is a decrease in consumption of a palatable sweet solution. The model has good predictive validity (behavioural changes are reversed by chronic treatment with a wide variety of antidepressants), face validity (almost all demonstrable symptoms of depression have been demonstrated), and construct validity (CMS causes a generalized decrease in responsiveness to rewards, comparable to anhedonia, the core symptom of the melancholic subtype of major depressive disorder). Overall, the CMS procedure appears to be at least as valid as any other animal model of depression. The procedure does, however, have two major drawbacks. One is the practical difficulty of carrying out CMS experiments, which are labour intensive, demanding of space, and of long duration. The other is that, while the procedure operates reliably in many laboratories, it can be difficult to establish, for reasons which remain unclear. However, once established, the CMS model can be used to study problems that are extremely difficult to address by other means.