OBJECTIVES: Use of objectively measured physical activity (PA) in older adults to assess relationship between PA and risk of all-causes mortality is scarce. This study evaluated the associations of PA based on accelerometry and a questionnaire with the risk of mortality among older adults from a city in Southern Brazil. DESIGN: A cohort study. SETTING: Urban area of Pelotas, Southern Brazil. PARTICIPANTS: A representative sample of older adults (≥60 y) from Pelotas, enrolled in 2014. MEASUREMENTS: Overall physical activity (mg), light physical activity (LPA), and moderate to vigorous physical activity (MVPA) were estimated by raw accelerometer data. The International Physical Activity Questionnaire estimated leisure time and commuting PA. Hazard ratios (excluding deaths in the first 6 mo) stratified by sex were estimated by Cox regression analysis considering adjustment for confounders. RESULTS: From the 1451 older adults interviewed in 2014, 145 died (10%) after a follow-up of an average 2.6 years. Men and women in the highest tertile of overall PA had on average a 77% and 92% lower risk of mortality than their less active counterparts (95% confidence interval [CI] = .06-.84 and 95% CI = .01-.65, respectively). The highest tertile of LPA was also related to a lower risk of mortality in individuals of both sexes (74% and 91% lower risk among men and women, respectively). MVPA statistically reduced the risk of mortality only among women (hazard ratio [HR] = .30 and HR = .07 in the second and third tertiles). Self-reported leisure-time PA was statistically associated with a lower risk of mortality only among men. Women in the highest tertiles of commuting PA showed a lower risk of mortality than those in the reference group. CONCLUSION: Accelerometry-based PA was associated with a lower risk of mortality among Brazilian older adults. Older individuals should practice any type of PA. J Am Geriatr Soc 68:137-146, 2019.