Objective: Investigate all-cause and cause-specific late mortality after childhood acute lymphoblastic leukemia (ALL) in a population-based Nordic cohort. Methods: From the cancer registries of Denmark, Finland, and Sweden, we identified 3765 five-year survivors of ALL, diagnosed before age 20 during 1971–2008. For each survivor, up to five matched comparison subjects were randomly selected from the general population (n = 18,323). Causes of death were classified as relapse related, health related, and external. Late mortality was evaluated by cumulative incidences of death from 5-year survival date. Mortality hazard ratios (HR) were evaluated with Cox proportional models. Results: Among the survivors, 315 deaths occurred during a median follow-up of 16 years from 5-year survival date (range 0–42). The majority were attributable to relapse (n = 224), followed by second neoplasm (n = 45). Cumulative incidence of all-cause late mortality at 15 years from diagnosis decreased gradually over treatment decades, from 14.4% (95% confidence interval [CI]: 11.6–17.2) for survivors diagnosed during 1971–1981, to 2.5% (95% CI: 1.3–3.7) for those diagnosed during 2002–2008. This was mainly attributable to a reduction in relapse-related deaths decreasing from 13.4% (95% CI: 10.7–16.1) for survivors diagnosed during 1971–1981 to 1.9% (95% CI: 0.9–2.8) for those diagnosed during 2002–2008. Health-related late mortality was low and did not change substantially across treatment decades. Compared to comparison subjects, all-cause mortality HR was 40 (95% CI: 26–61) 5–9 years from diagnosis, and 4.4 (95% CI: 3.4–5.6) ≥10 years from diagnosis. Conclusions: Survivors of ALL have higher late mortality than population comparison subjects. Among the survivors, there was a temporal reduction in risk of death from relapse, without increments in health-related death.