Aim: Childhood cancer survivors are at risk of both somatic and mental late effects, but large population-based studies of depression are lacking. Methods: Risk of antidepressant use was evaluated in a population-based cohort of 5452 Danish children treated for cancer in 1975-2009 by linkage to the National Prescription Drug Database, which worldwide is the oldest nationwide registry of prescription medication. Hazard ratios (HRs) for antidepressant use were estimated in a Cox proportional hazards model stratified on sex, with population comparisons as referents. Results: Overall, childhood cancer survivors were at increased risk of having antidepressants prescribed (HR, 1.4; 95% confidence interval (CI), 1.3-1.5). The excess absolute risk of antidepressant use was 2.5 per 1000 person-years (95% CI, 1.7-3.3), equivalent to an excess of 2.5 survivors for every 100 survivors followed for 10 years. Increased HRs of 30-50% were seen for survivors of cancers of all main groups (haematological malignancies, central nervous system (CNS) and solid tumors); the highest risk was among children treated with haematopoietic stem cell transplantation (HR, 1.9; 95% CI, 1.2-3.1). Our data suggested that the risk was most pronounced for children treated in the most recent calendar periods (test for interaction between cancer and calendar periods: P < 0.001), especially for survivors of haematological cancers (P = 0.007). Interaction analysis of the effect of parental socioeconomic position and psychiatric disease on the association between childhood cancer and antidepressant use indicated no modifying effect. Conclusion Childhood cancer survivors should be followed-up for depression. Our results indicate an increasing need for follow-up especially in survivors treated by more recent, intensive anticancer treatment.