OBJECTIVES: To evaluate the effectiveness of a multifactorial fall and injury prevention program in older people with higher and lower levels of cognition. DESIGN: A preplanned subgroup comparison of the effectiveness of a cluster-randomized, nonblinded, usual-care, controlled trial. SETTING: Nine residential facilities in Umeå, Sweden. PARTICIPANTS: All consenting residents living in the facilities, aged 65 and older, who could be assessed using the Mini-Mental State Examination (MMSE; n = 378). An MMSE score of 19 was used to divide the sample into one group with lower and one with higher level of cognition. The lower MMSE group was older (mean +/- standard deviation = 83.9 +/- 5.8 vs 82.2 +/- 7.5) and more functionally impaired (Barthel Index, median (interquartile range) 11 (6-15) vs 17 (13-18)) and had a higher risk of falling (64% vs 36%) than the higher MMSE group. INTERVENTION: A multifactorial fall prevention program comprising staff education, environmental adjustment, exercise, drug review, aids, hip protectors, and postfall problem-solving conferences. MEASUREMENTS: The number of falls, time to first fall, and number of injuries were evaluated and compared by study group (intervention vs control) and by MMSE group. RESULTS: A significant intervention effect on falls appeared in the higher MMSE group but not in the lower MMSE group (adjusted incidence rates ratio of falls P =.016 and P =.121 and adjusted hazard ratio P