86 cognitively-impaired elderly patients were randomized to undergo either a multi-modal exercise program, or conventional care (e.g. participation in social activities) to test the effect a multi-modal exercise program has on balance, functional mobility, and fall risk in this population. Results demonstrated a significant improvement in the Performance Oriented Mobility Assessment scale at final follow up (12 months) for both balance and gait scores in the exercise group, but not in the control. Timed Up and Go (TUG) was also significantly improved in the exercise group. No between group difference was found when looking at fall incidences at 6 or 12 months.