Background & aims Older diabetic people are at risk of cognitive decline. Vitamin B12 deficiency in older people is associated with cognitive impairment and Alzheimer's disease. Vitamin B12 deficiency may therefore contribute to cognitive decline in older diabetic people. We therefore performed a randomized placebo-controlled trial of vitamin B12 supplementation to prevent cognitive decline in older diabetic people with mild vitamin B12 deficiency. Methods 271 diabetic non-demented outpatients aged 70 years or older with plasma vitamin B12 150–300 pmol/L in outpatient clinics were randomly assigned to take either methylcobalamin 1000 μg or two similar looking placebo tablets once daily for 27 months. All subjects were followed up at 9 monthly intervals. The primary outcome is cognitive decline as defined by an increase in clinical dementia rating scale (CDR) global score. The secondary outcomes included Neuropsychological Test Battery (NTB) z-scores, serum methymalonic acid (MMA) and homocysteine. Results The subjects in the trial groups were well matched in clinical characteristics, except that active intervention group had more smokers. 46.5% and 74.1% had elevated serum methymalonic acid (≥0.21 μmol/L) and homocysteine (≥13 μmol/L) respectively. 44% of the subjects had CDR score of 0.5 suggesting questionable dementia. At month 9 and 27, serum MMA and homocysteine was significantly reduced in the active treatment group, when compared with placebo group. (P < 0.0001, student t test) At month 27, there was no significant group difference in changes in CDR or NTB z-scores. Exclusion of smokers did not alter the results. Subgroup analysis of high MMSE and serum MMA showed similar results. Conclusion Vitamin B12 supplementation did not prevent cognitive decline in older diabetic patients with borderline vitamin B12 status. Clinical trial registration ClinicalTrials.gov: NCT02457507.