To estimate the number of fallers and risk factors for falls in a cohort with dementia, we did a secondary analysis of a cluster-randomized controlled trial (NutriAlz) in 11 outpatient and day care centers in Catalonia (Spain) including 626 community-dwelling patients with dementia, followed for 12 months. Participants’ characteristics were assessed at baseline, at 6 and 12 months [fall in the earlier 6 mo, anthropometric data, comorbidities, Mini-Mental State Examination, Clinical Dementia Rating, Basic Activities of Daily Living (BADL), Instrumental Activities of Daily Living, Neuropsychiatric Inventory Questionnaire, Zarit Caregiver Burden Interview and Mini-Nutritional Assessment]. Multivariate logistic regression models and generalized linear models were used to explore risk factors for falls and changes in health and function. Two hundred twenty-three participants fell during the 12 months follow-up (35.62%). Risk factors identified for falls were age (odds ratio (OR)=1.03, 95% confidence interval (CI), 1.00-1.05), BADL (OR=1.18, 95% CI, 1.05-1.32), and earlier fall (OR=2.30, 95% CI, 1.57-3.35). Fallers had worse health than nonfallers, and their dependence increased significantly more in BADL during the study, compared with nonfallers. Dependence in BADL is a risk factor and a consequence of falls; interventions aimed at preventing falls in dementia patients could promote autonomy in BADL and slow its decline.