This study aimed to investigate the effects of semi-immersive virtual reality-based cognitive training (VRCT) combined with locomotor activity on cognitive function, balance, and gait ability in older adults. Eighteen community-dwelling older adults participated in this study. Subjects who met the selection criteria were assigned to an experimental group ( = 9) and a control group ( = 9). The experimental group received VRCT combined with locomotor activity for 30 min a day, three times a week, for 6 weeks. The control group received tabletop activity-based cognitive training for the same amount of time. Before and after the training, the Korean Mini-Mental State Examination (K-MMSE), Trail Making Test (TMT; A and B), and Digit Span Test (DST; forward and backward) were used to evaluate cognitive function; and the Timed Up and Go (TUG) test and 10-m Walking Test (10MWT) were used to evaluate the improvement in the balance and gait ability parameters. After the intervention, the experimental group showed a significantly greater improvement in the TMT-A ( = 0.045) and DST-backward ( = 0.012) scores compared with the control group. Regarding the gait ability variable, the experimental group showed a significant improvement in the 10MWT test ( = 0.001). This study confirmed that semi-immersive VRCT combined with locomotor activity is useful for improving cognitive function and gait ability in older adults. Therefore, VRCT combined with locomotor activity can be used as a simultaneous intervention for cognitive rehabilitation and functional capacity improvement in older adults.