BackgroundMass incarceration has collateral consequences for community health, which are reflected in county-level health indicators, including county mortality rates. County jail incarceration rates are associated with all-cause mortality rates in the USA. We assessed the causes of death that drive the relationship between county-level jail incarceration and mortality.MethodsIn this retrospective, longitudinal study, we assessed the association between county-level jail incarceration rates and county-level cause-specific mortality using county jail incarceration data (1987–2017) for 1094 counties in the USA obtained from the Vera Institute of Justice and cause-specific mortality data for individuals younger than 75 years in the total county population (1988–2018) obtained from the US National Vital Statistics System. We fitted quasi-Poisson models for nine common causes of death (cerebrovascular disease, chronic lower respiratory disease, diabetes, heart disease, infectious disease, malignant neoplasm, substance use, suicide, and unintentional injury) with county fixed effects, controlling for all unmeasured stable county characteristics and measured time-varying confounders (county median age, county poverty rate, county percentage of Black residents, county crime rate, county unemployment rate, and state incarceration rate). We lagged county jail incarceration rates by 1 year to assess the short-term, by 5 years to assess the medium-term, and by 10 years to assess the long-term associations of jail incarceration with premature mortality.FindingsA 1 per 1000 within-county increase in jail incarceration rate was associated with a 6·5% increase in mortality from infectious diseases (risk ratio 1·065, 95% CI 1·061–1·070), a 4·9% increase in mortality from chronic lower respiratory disease (1·049, 1·045–1·052), a 2·6% increase in mortality induced from substance use (1·026, 1·020–1·032), a 2·5% increase in suicide mortality (1·025, 1·020–1·029), and smaller increases in mortality from heart disease (1·021, 1·019–1·023), unintentional injury (1·015, 1·011–1·018), malignant neoplasm (1·014, 1·013–1·016), diabetes (1·013, 1·009–1·018), and cerebrovascular disease (1·010, 1·007–1·013) after 1 year. Associations between jail incarceration and cause-specific mortality rates weakened as time lags increased, but to a greater extent for causes of death with generally shorter latency periods (infectious disease and suicide) than for those with generally longer latency periods (heart disease, malignant neoplasm, and cerebrovascular disease).InterpretationJail incarceration rates are potential drivers of many causes of death in US counties. Jail incarceration can be harmful not only to the health of individuals who are incarcerated, but also to public health more broadly. Our findings suggest important points of intervention, including disinvestment from carceral systems and investment in social and public health services, such as community-based treatment of substance-use disorders.