AIMS: Gait characteristics and balance are altered in diabetic individuals. Little is known about possible treatment strategies. This study evaluated the effect of a specific training program on diabetic patients' gait. METHODS: A randomized controlled trial (N=71) with an intervention (IG) (N=35), and control group (CG) (N=36). The intervention consisted of physiotherapeutic group training including gait and balance exercises with function-oriented strengthening. Controls received no treatment. RESULTS: After intervention the IG increased their habitual walking speed by 0.149ms(-1) (0.54kmh(-1)) on tarred terrain and by 0.169ms(-1) (0.61kmh(-1)) on the cobblestones. This significant treatment effect (p<0.001) decreased slightly at the six-month follow-up, but remained significant (p<0.001). In a similar manner, significant improvement was observed for cadence, gait cycle time and stance time on both terrains. All outcomes except stance time on the tarred terrain remained significant at the six-month follow-up. No significant effect was observed for stride length and the coefficient of variation of gait cycle time (on either surface) at the corrected significance level of p<0.004. CG patients' parameters all remained unchanged or progressively deteriorated compared to baseline values. DISCUSSION: Cadence contributed 80%, whereas stride length only contributed 20% to the change of gait velocity. This may be due to the treatment or to diabetic patients' potential to regulate their cadence and stride length. CONCLUSION: A specific training program can improve diabetic patients' gait in a real life environment. A challenging environment highlights treatment effects on patients' gait better than an evenly tarred surface.