Introduction Previous evidence has shown significant effects of exercise, cognitive and dual-task training for improving cognition in healthy cohorts. The effects of these types of interventions in type 2 diabetes mellitus is unclear. The aim of this research was to systematically review evidence, and estimate the effect, of exercise, cognitive, and dual-task interventions on cognition in type 2 diabetes mellitus. Method Electronic databases including PubMed, EMBASE, CINAHL, PsycINFO, SPORTDiscus, and MEDLINE were searched for ongoing and completed interventional trials investigating the effect of either an exercise, cognitive or dual-task intervention on cognition in type 2 diabetes mellitus. Results Nine trials met the inclusion criteria–one dual-task, two cognitive, and six exercise. Meta-analyses of exercise trials showed no significant effects of exercise on measures of executive function (Stroop task, SMD = -0.31, 95% CI -0.71–0.09, P = 0.13, trail making test part A SMD = 0.28, 95% CI -0.20–0.77 P = 0.25, trail making test part B SMD = -0.15, 95% CI -0.64–0.34 P = 0.54, digit symbol SMD = 0.09, 95% CI -0.39–0.57 P = 0.72), and memory (immediate memory SMD = 0.20, 95% CI -0.28–0.69, P = 0.41 and delayed memory SMD = -0.06, 95% CI -0.55–0.42, P = 0.80). A meta-analysis could not be conducted using cognitive or dual-task data, but individual trials did report a favourable effect of interventions on cognition. Risk of bias was considered moderate to high for the majority of included trials. Conclusions Meta-analyses of exercise trials identified a small effect size (0.31), which whilst not significant warrants further investigation. Larger and more robust trials are needed that report evidence using appropriate reporting guidelines (e.g. CONSORT) to increase confidence in the validity of results.