OBJECTIVES: To determine the incidence of fecal incontinence (FI) in community-dwelling older adults and identify risk factors associated with incident FI. DESIGN: Planned secondary analysis of a longitudinal, population-based cohort study. SETTING: Three rural and two urban Alabama counties (in-home assessments 2000-2005). PARTICIPANTS: Stratified random sample of 1,000 Medicare beneficiaries: 25% African-American men, 25% white men, 25% African-American women, 25% white women, aged 65 and older. Eligible participants for this analysis were continent at baseline and community-dwelling 4 years later (n=557). MEASUREMENTS: FI was defined as any loss of control of bowels occurring during the previous year. Independent variables were sociodemographics, Charlson comorbidity counts, self-reported bowel symptoms (chronic diarrhea and constipation), depression, and body mass index (BMI). Multivariable logistic regression models were constructed using incident FI as the dependent variable. RESULTS: The incidence rate of FI at 4 years was 17% (95% confidence interval (CI)=13.7-20.1), with 6% developing FI at least monthly (95% CI=4.0-8.3). White women were more likely to have incident FI (22%) than African-American women (13%, P=.04); no racial differences were observed in men. Controlling for age, comorbidity count, and BMI, significant independent risk factors for incident FI in women were white race, depression, chronic diarrhea, and urinary incontinence (UI). UI was the only significant risk factor for incident FI in men. CONCLUSION: The occurrence of new FI is common in men and women aged 65 and older, with a 17% incidence rate over 4 years. FI and UI may share common pathophysiologic mechanisms and need regular assessment in older adults.