Despite the importance of regulating working hours for workers' health and maintaining labour productivity, the literature lacks credible causal estimates on the impact of reduced working hours. We provide new evidence for the causal effect of shorter workweeks on mortality using full population register data, exploiting a nationwide policy in Sweden that reduced the weekly working hours from 55 to 48 hours for certain occupations only in 1920. Using difference-in-differences and event-study models, we show that lower working hours decreased mortality by around 15% over the first six years. We identify several mechanisms behind this effect: the policy led to fewer workplace accidents, a decline in work-related disability, and a reduction in sick days taken by employees. Causal forest estimators indicate particularly strong effects for older workers. Our results imply that many lives could be saved worldwide by reducing long working hours for labour-intensive occupations.