There are several animal models of type 2 diabetes mellitus induction but the comparison between models is scarce. Food restriction generates benefits, such as reducing oxidative stress, but there are few studies on its effects on diabetes. The objective of this study is to evaluate the differences in physiological and biochemical parameters between diabetes models and their responses to food restriction. For this, 30 male Wistar rats were distributed in 3 groups (n = 10/group): control (C); diabetes with streptozotocin and cafeteria-style diet (DE); and diabetes with streptozotocin and nicotinamide (DN), all treated for two months (pre-food restriction period). Then, the 3 groups were subdivided into 6, generating the groups CC (control), CCR (control+food restriction), DEC (diabetic+standard diet), DER (diabetic+food restriction), DNC (diabetic+standard diet) and DNR (diabetic+food restriction), treated for an additional two months (food restriction period). The food restriction (FR) used was 50% of the average daily dietary intake of group C. Throughout the treatment, physiological and biochemical parameters were evaluated. At the end of the treatment, serum biochemical parameters, oxidative stress and insulin were evaluated. Both diabetic models produced hyperglycemia, polyphagia, polydipsia, insulin resistance, high fructosamine, hepatic damage and reduced insulin, although only DE presented human diabetes-like alterations, such as dyslipidemia and neuropathy symptoms. Both DEC and DNC diabetic groups presented higher levels of protein carbonyl groups associated to lower antioxidant capacity in the plasma. FR promoted improvement of glycemia in DNR, lipid profile in DER, and insulin resistance and hepatic damage in both diabetes models. FR also reduced the protein carbonyl groups of both DER and DNR diabetic groups, but the antioxidant capacity was improved only in the plasma of DER group. It is concluded that FR is beneficial for diabetes but should be used in conjunction with other therapies.