Regular exercise has been recommended for diabetes patients for many years; however, it is not widely used clinically. This may be because of high costs, lack of reimbursement, low compliance and/or absence of proper infrastructure. Alternatively, structured exercise therapy may be underutilised because current guidelines do not include detailed information on the preferred type and intensity of exercise that should be applied to maximise the benefits of exercise for different subgroups of patients with type 2 diabetes. Based on available evidence and our own clinical research experience this article proposes that exercise therapy in type 2 diabetes might be more effective if (1) cardiac rehabilitation programmes served as a model for 'pre-cardiac diabetes rehabilitation'; (2) resistance exercise were prescribed for sarcopenic or severely deconditioned type 2 diabetes patients; and (3) a multidisciplinary approach and continued exercise training under personal supervision became standard therapy. Nevertheless, more clinical research is warranted to establish the efficacy of an approach that takes into account type 2 diabetes subpopulations at different stages of the disease and with different levels of comorbidity.