Self-rated health is a common assessment used in epidemiological research and an independent predictor of morbidity and mortality. We investigate if a single measure of self-rated health in late adolescence predict mortality between 46 and 70 years of age. This study was based on 47 286 Swedish men that conscribed 1969–1970 at age 18–20 and that were still alive in 1997. Self-rated health and data on potential explanatory factors (psychological factors, health status markers and health behaviors) were collected at conscription. Adult socioeconomic position in 1990 was derived from registries. Death and cause of death (cancer, cardiovascular disease, violent and alcohol abuse related disorders) were derived from the Causes of Death Register between 1997 and 2019. Conscripts that rated their health as fair or poor/very poor had significantly higher hazard of all-cause mortality than conscript that rated their health as very good (HRfair = 1.27, 95%CI:1.18–1.37 and HRpoor = 1.25, 95%CI:1.11–1.41) and disease-specific mortality. Adjusting for all explanatory factors attenuated the risk estimates by 9–100%. In conclusion, poor self-rated health reported in late adolescence predicts all-cause and disease-specific mortality between the ages of 46 and 70 years. Psychological factors and health behaviors measured at conscription may serve as potential explanatory factors underlying the predictive ability of self-rated health in late adolescence.