Background: AYA with cancer are a vulnerable sub-population at risk of adverse mental health outcomes during and after cancer treatment. Tools to identify AYA at highest risk are required to guide screening and interventions. In a population-based cohort of AYA with cancer, we determined whether self-reported symptoms were associated with subsequent short- and long-term severe mental health events (SMHE). Methods: All Ontario, Canada AYA diagnosed with cancer aged 15-29 between 2010-2018 were identified and linked to healthcare databases, including one capturing self-reported Edmonton Symptom Assessment System (ESAS) scores at cancer-related visits. Scores for depression, anxiety, and poor well-being were categorized as not measured, mild (0-3), moderate (4-6), or severe (7-9). SMHE were defined as emergency room visits or hospitalizations for mental health reasons. First, we used Cox proportional hazard models to determine the association of ESAS scores (time-varying variable) with subsequent SMHE. Second, among 5-year survivors, we determined the association of maximum ESAS score within the first year of diagnosis with long-term SMHE (i.e. starting at 5 years from cancer diagnosis). All analyses were adjusted for patient and disease variables, including mental healthcare use prior to cancer diagnosis. Results: 5,435 AYA met inclusion criteria. Median age at cancer diagnosis was 25 years [interquartile range 22-27]. Hematologic cancers were most common (1,748; 32.2%). Symptom severity was associated with subsequent SMHE risk. For example, AYA reporting severe anxiety were at more than three-fold higher risk of SMHE compared to those reporting mild anxiety [adjusted hazard ratio (aHR) 3.6, 95^th confidence interval (CI) 1.9-6.7; p < 0.001]. Similar risk was seen among AYA reporting severe vs. mild depression (aHR 3.5, 1.7-7.3; p < 0.001). Among 3,518 (64.7%) 5-year survivors, symptom severity also predicted long-term SMHE. For example, starting at 5 years post cancer diagnosis, the subsequent 3-year cumulative incidence of a SMHE among those who reported severe depression at any time during the first year post cancer diagnosis was 10.5% (95CI 6.9-15.9) compared to 2.4% (95CI 1.7-3.3) among those who only reported mild depression (aHR 3.0, 95CI 1.8-4.9; p < 0.0001). Similar results were seen pertaining to severe anxiety and severe impact on well-being. AYA endorsing severe anxiety represented 13.1% of the cohort but accounted for 25.8% of AYA experiencing SMHEs during the first three years of survivorship. Conclusions: Systematic symptom screening in the first year after cancer diagnosis identifies a proportion of AYA at high risk of both short and long-term SMHE who may benefit from targeted screening and interventions. Future work will determine whether interventions during cancer treatment mitigate this risk.