Preclinical biomedical models are a fundamental tool to improve the knowledge and management of diseases, particularly in diabetes mellitus (DM) since, currently, the pathophysiological and molecular mechanisms involved in its development are not fully clarified, and there is no treatment to cure DM. This review will focus on the features, advantages and limitations of some of the most used DM models in rats, such as the spontaneous models: Bio-Breeding Diabetes-Prone (BB-DP) and LEW.1AR1-iddm, as representative models of type 1 DM (DM-1); the Zucker diabetic fatty (ZDF) and Goto-kakizaki (GK) rats, as representative models of type 2 DM (DM-2); and other models induced by surgical, dietary and pharmacological-alloxan and streptozotocin-procedures. Given the variety of DM models in rats, as well as the non-uniformity in the protocols and the absence of all the manifestation of the long-term multifactorial complications of DM in humans, the researchers must choose the one that best suits the final objectives of the study. These circumstances, added to the fact that most of the experimental research in the literature is focused on the study of the early phase of DM, makes it necessary to develop long-term studies closer to DM in humans. In this review, a recently published rat DM model induced by streptozotocin injection with chronic exogenous administration of insulin to reduce hyperglycaemia has also been included in an attempt to mimic the chronic phase of DM in humans.