Many smokers do not quit but instead reduce the number of cigarettes they smoke per day (CPD) over their lifetime. Yet the associations of such changes in CPD with health risks are unclear. We examined the association of changes in CPD with subsequent death in the period 2004-2011 among 253,947 participants of the National Institutes of Health-AARP Diet and Health Study. Using a questionnaire assessing responders' history of smoking cigarettes, we identified cigarette smokers who quit, decreased, maintained, or increased their CPD between ages 25-29 and 50-59 years. Hazard ratios and 95% confidence intervals were obtained from multivariable adjusted Cox proportional hazards regression models. Relative to never smokers, smokers who maintained a consistent CPD had 2.93 times (95% confidence interval (CI): 2.82, 3.05) higher all-cause mortality risk, and participants who increased their CPD had still higher risk (hazard ratio (HR) = 3.37, 95% CI: 3.23, 3.52). Death risk was lower among participants who decreased their CPD (HR = 2.38, 95% CI: 2.25, 2.52) or quit smoking (for quitting between ages 30 and 39 years, HR = 1.32, 95% CI: 1.25, 1.39). Similar patterns were observed for smoking-related causes of death, with particularly strong associations for lung cancer and respiratory disease. Reductions in CPD over the lifetime meaningfully decreased death risk; however, cessation provided a larger benefit than even large declines in CPD.