Background: Few studies have adequately addressed long-term survival ('20 years from diagnosis) among survivors of adolescent and young adult (AYA) cancers. Methods: In this retrospective, population-based cohort study in a US integrated health care system, the authors examined cause-specific mortality in 2-year survivors of AYA cancers (patients aged 15-39 years who were diagnosed between 1990 and 2012; N = 10,574) matched (by age, sex, and calendar year) to individuals without cancer (N = 136,683) to determine whether mortality rates changed over time. Incidence rate ratios (IRRs) for mortality were estimated using multivariable Poisson regression. A multivariable Cox model was used to examine predictors of cause-specific mortality among AYA cancer survivors. Results: Through December 31, 2014, 1352 deaths were observed among AYA cancer survivors, yielding an overall survival rate of 78.5% at 25 years after diagnosis. Overall, AYA cancer survivors were at 10.4-fold increased risk for death (95% CI, 9.7-fold to 11.2-fold increased risk for death) compared with the matched noncancer cohort, and this risk remained elevated at '20 years after diagnosis (IRR, 2.9; 95% CI, 2.0-4.3). The absolute excess risk for death from any cause was 12.7 per 1000 person-years (95% CI, 11.9-13.4 per 1000 person-years). Starting at 15 years after diagnosis, the incidence of second cancer-related mortality exceeded the rate of recurrence-related mortality, and similar trends were observed for deaths from other health-related conditions. The 8-year cumulative incidence of mortality declined over time (before 2000, 12.6%; 2000-2006, 10.1%; after 2006, 7.3%; P '.001), largely because of declines in recurrence-related mortality. Age, sex, race/ethnicity, cancer stage at diagnosis, and cancer treatment predicted cause-specific mortality. Conclusions: The current data highlight the need for specialized, long-term follow-up care for AYA cancer survivors.