BACKGROUND AND PURPOSE: Individuals with Alzheimer disease (IwAD) experience more frequent and more injurious falls than their cognitively intact peers. Evidence of balance and gait dysfunction is observed earlier in the course of Alzheimer disease (AD) than once believed. Balance training has been demonstrated to be effective in improving balance and decreasing falls in cognitively intact older adults but is not well studied in IwAD. This study was designed to analyze the effects of a group balance training program on balance and falls in IwAD. The program was developed specifically for IwAD, with explicit guidelines for communication/interaction and deliberate structure of training sessions catered to the motor learning needs of IwAD. DESIGN: This prospective, quasi-experimental, pretest-posttest design study describes the effects of a balance training program for a cohort of IwAD. METHODS: Thirty IwAD were recruited from 3 adult day health centers; 22 completed at least 1 posttest session. Participants were tested with Berg Balance Scale (BBS), Timed Up and Go (TUG), Self-Selected Gait Speed (SSGS), Fast Gait Speed (FGS), and Mini-Mental Status Examination (MMSE) immediately before and after the 3-month intervention and again 3 months later. Group training was held at the adult day health centers for 45 minutes, twice per week. Sessions were characterized by massed, constant, and blocked practice of functional, relevant activities with considerable repetition. Ratio of participant to staff member never exceeded 3:1. Physical therapist staff members assured that participants were up on their feet the majority of each session and were individually challenged as much as possible. RESULTS: Repeated-measures analysis of variance (ANOVA) for BBS was significant (F = 15.04; df = 1.67/28.40; P = .000) with post hoc tests, revealing improvement between pretest and immediate posttest (P = .000) and decline in performance between immediate and 3-month posttest (P = .012). Repeated-measures ANOVA posttest for MMSE was significant (F = 5.12; df = 1.73/22.53; P = .018) with post hoc tests, showing no change in MMSE between pretest and immediate posttest but decline in MMSE when comparing immediate posttest with 3-month posttest (P = .038) and pretest with 3-month posttest (P = .019). Repeated-measures ANOVA for TUG, FGS, and SSGS were not significant. Immediate effects of the intervention as assessed by the a priori paired t tests (comparing pre- and immediate posttest data) revealed significant improvement in BBS (t = -7.010; df = 20; P = .000), TUG (t = 3.103; df = 20; P = .006), and FGS (t = -2.115; df = 19; P = .048), but not in SSGS (t = -1.456; df = 20; P = .161). DISCUSSION AND CONCLUSION: The 3-month group balance training intervention designed specifically for IwAD was effective in improving balance and mobility, as evidenced by improved BBS and TUG performances. Cognition did not decline during the course of the intervention but did decline following the intervention, suggesting a possible protective effect. Given the promising findings, a larger-scale controlled study is warranted.