Background: Dementia is an increasingly important public health concern globally. Older adults with cognitive impairments are subjected to poorer balance and movement coordination, lower physical activity level, higher risk of falls and fractures than elderly people without disability. Purpose: A systematic review was undertaken to evaluate the effect of exercise on physical health in people with cognitive impairments. Methods: Electronic search for publications concerning the effect of physical exercise on people with cognitive impairments were conducted using Medline, CINAHL, Cochrane, Psyco INFO and Pubmed. The inclusion criteria were: (1) randomized controlled trials; (2) investigating the effect of exercise on people whose cognitive impairment was the primary diagnosis (e.g., dementia, mild cognitive impairment); (3) published in English. Exclusion criteria were (1) reports published as conference proceedings; (2) reports in books. The last search was performed on August 22, 2014. Data were extracted from articles that fulfilled the selection criteria. The overall effects of exercise on physical functioning were then synthesized and analyzed. Meta analysis was performed if five similar studies adopted comparable outcome measures. The PEDro score was used to assess the methodological quality of each selected study (PEDro score 9-10: excellent; 6-8: good; 4-5: fair; <4: poor). Results: A total of 9635 records were identified through database searching. Forty-four papers (37 trials) fulfilled the selection criteria and were included in the review. The sample size of the studies ranged from 16 to 389. The mean age of the participants was between 69 and 89. Most studies involved people with early stage of cognitive impairment. Different forms of physical exercise training were studied, including functional training, muscle strength training, physiotherapy, walking exercise, proprioception and balance training, flexibility training, aerobic training, dance and movement therapy, dual-task training, and integrated exercise training. For methodological quality, 2, 11, 29 and 2 articles were rated as poor, fair, good and excellent respectively. Our meta-analyses revealed that exercise had significant beneficial effects on lower limb strength (Standardized mean difference (SMD) = 0.72, 95%Confidence Interval (CI) = 0.16-1.29; p = 0.01), balance performance (SMD = 0.78- 0.82, 95%CI = 0.46-1.16; p < 0.01), Timed-Up-and-Go test performance (SMD = 0.20, 95%CI = 0.04-0.36; p = 0.01), walking speed (SMD = 0.52-0.53, 95%CI = 0.03-1.02; p = 0.04), step length (SMD = 0.97, 95%CI = 0.07-1.87; p = 0.03), endurance (SMD = 0.43-0.46, 95%CI = 0.02-0.86; p = 0.02-0.04), activities of daily living (SMD = 0.56-0.70, 95%CI = 0.21-1.06; p < 0.01) and quality of life (SMD = 0.17-0.20, 95%CI = 0.00-0.36; p = 0.02-0.04) in individuals with cognitive impairments. The number of good-quality trials investigating effect of exercise on flexibility, fall rate and dual-task ability was small and meta-analysis was not justified. The long-term effect of exercise was also rarely investigated. Exercise adherences are generally acceptable across trials with rare adverse effects. Conclusion(s): Physical exercise training is feasible in people with cognitive impairment, particularly in those in early stage, and can improve various aspects of physical health and quality of life. Implications: Physical exercise intervention should be recommended for people with cognitive impairment in order to maintain or improve physical function. Further studies are needed to examine the long-term effects of physical exercise.