ObjectiveTo test if patients recovering from COVID-19 are at increased risk of mental morbidities and to what extent such risk is exacerbated by illness severity.DesignPopulation-based cross-sectional study.SettingIceland.ParticipantsA total of 22 861 individuals were recruited through invitations to existing nationwide cohorts and a social media campaign from 24 April to 22 July 2020, of which 373 were patients recovering from COVID-19.Main outcome measuresSymptoms of depression (Patient Health Questionnaire), anxiety (General Anxiety Disorder Scale) and posttraumatic stress disorder (PTSD; modified Primary Care PTSD Screen for DSM-5) above screening thresholds. Adjusting for multiple covariates and comorbidities, multivariable Poisson regression was used to assess the association between COVID-19 severity and mental morbidities.ResultsCompared with individuals without a diagnosis of COVID-19, patients recovering from COVID-19 had increased risk of depression (22.1% vs 16.2%; adjusted relative risk (aRR) 1.48, 95% CI 1.20 to 1.82) and PTSD (19.5% vs 15.6%; aRR 1.38, 95% CI 1.09 to 1.75) but not anxiety (13.1% vs 11.3%; aRR 1.24, 95% CI 0.93 to 1.64). Elevated relative risks were limited to patients recovering from COVID-19 that were 40 years or older and were particularly high among individuals with university education. Among patients recovering from COVID-19, symptoms of depression were particularly common among those in the highest, compared with the lowest tertile of influenza-like symptom burden (47.1% vs 5.8%; aRR 6.42, 95% CI 2.77 to 14.87), among patients confined to bed for 7 days or longer compared with those never confined to bed (33.3% vs 10.9%; aRR 3.67, 95% CI 1.97 to 6.86) and among patients hospitalised for COVID-19 compared with those never admitted to hospital (48.1% vs 19.9%; aRR 2.72, 95% CI 1.67 to 4.44).ConclusionsSevere disease course is associated with increased risk of depression and PTSD among patients recovering from COVID-19.