This study aims to determine the effects of transtheoretical model-based walking exercise training and follow-up on improving exercise behavior and metabolic control in patients with type 2 diabetes. This randomized controlled trial was conducted as a pre-test and post-test experimental model with 76 intervention and 76 control individuals. The intervention group received the transtheoretical model-based exercise training. Data were collected using patient identification form, transtheoretical model scales, pedometer and metabolic outcomes. Data were analyzed by Chi square, two-way ANOVA, Mauchly's, Greenhouse-Geisser, Friedman, and McNamer test. The overall score means for the exercise change processes, decision-making balance and self-efficacy scales increased compared to the pre-test (p < .001). The mean HDL increased from 41.39 +/- 10.35 to 49.18 +/- 11.58, and average number of steps per day increased from 3264.31 +/- 1933.03 to 5639.37 +/- 2317.01. Consequently, this difference between the groups was significant (p < .05).