The US physical activity (PA) recommendations were based primarily on studies in which self-reported data were used. Studies that include accelerometer-assessed PA and sedentary behavior can contribute to these recommendations. In the present study, we explored the associations of PA and sedentary behavior with all-cause and cardiovascular disease (CVD) mortality in a nationally representative sample. Among the 2003-2006 National Health and Nutrition Examination Survey cohort, 3,809 adults 40 years of age or older wore an accelerometer for 1 week and self-reported their PA levels. Mortality data were verified through 2011, with an average of 6.7 years of follow-up. We used Cox proportional hazards models to obtain adjusted hazard ratios and 95% confidence intervals. After excluding the first 2 years, there were 337 deaths (32% or 107 of which were attributable to CVD). Having higher accelerometer-assessed average counts per minute was associated with lower all-cause mortality risk: When compared with the first quartile, the adjusted hazard ratio was 0.37 (95% confidence interval: 0.23, 0.59) for the fourth quartile, 0.39 (95% confidence interval: 0.27, 0.57) for the third quartile, and 0.60 (95% confidence interval: 0.45, 0.80) second quartile. Results were similar for CVD mortality. Lower all-cause and CVD mortality risks were also generally observed for persons with higher accelerometer-assessed moderate and moderate-to-vigorous PA levels and for self-reported moderate-to-vigorous leisure, household and total activities, as well as for meeting PA recommendations. Accelerometer-assessed sedentary behavior was generally not associated with all-cause or CVD mortality in fully adjusted models. These findings support the national PA recommendations to reduce mortality.