Adults with type 2 diabetes mellitus have an increased risk for dementia. Therefore, we proposed an intervention called the Square-stepping exercise (SSE) program to mitigate this risk. Our primary aim was to determine the feasibility of SSE in adults with type 2 diabetes and self-reported cognitive complaints. Our secondary aim was to determine whether 24 weeks of SSE improved cognition. Our tertiary aim was to determine whether SSE improved antisaccade reaction time, which is a measure of executive-related oculomotor control. Adults >49 years with type 2 diabetes and self-reported cognitive complaints were randomized to an SSE group (2×/week for 24 weeks of SSE) or a control group. Feasibility was assessed by recruitment and attendance. Participants were assessed at baseline, after 12 weeks and after 24 weeks for global cognitive function, memory, planning, reasoning and concentration via a computer-based cognitive battery (Cambridge Brain Sciences) and antisaccade reaction time (at baseline and 24 weeks). Participants in the SSE group were (mean [SD]): 65.9 (5.2) years old; 33% female; body mass index 33.3 kg/m2 (4.8) (n=12). Participants in the control group were 71.2 (6.9) years old; 31% female; body mass index 31.9 kg/m2 (4.6) (n=13). Over 24 weeks, attendance was 70.2% (SD 17.2) for 4/12 participants. There were 4 withdrawals and 1 adverse event. There were no differences in global cognitive functioning. The SSE group improved in planning domain change scores between 12 and 24 weeks (F=5.8, p=0.03, ηp2=0.28) compared to controls. In the SSE group, we found a nonsignificant improvement in antisaccade reaction time of 38 ms (SD 16), n=2, compared to 9 ms (SD 45) in the control group, n=8. SSE should be evaluated further to improve its feasibility in older adults with type 2 diabetes. This study provides preliminary evidence that SSE improves executive function in adults with type 2 diabetes and self-reported cognitive complaints.