Abstract Background This study aims to investigate how pre-existing health status and social background contribute to excess mortality during the COVID-19 crisis in Belgium. Methods The study population consists of 1.4 million adult members of Solidaris, the second largest health insurance fund in Belgium. Pre-existing health status was identified using health care reimbursement data, including medication use. Social characteristics included financial support, migrant background, and household composition. Excess mortality during the COVID-19 crisis was measured as the relative difference between all-cause mortality in 2020 or 2021 versus the pooled yearly estimates in 2015-2019. Yearly age- and sex-adjusted mortality rate ratios (MRR) were estimated using Poisson regression to study potential differences by health status and social factors. Results Compared with healthy persons, all investigated health problems were associated with increased mortality, with most notably an MRR of 8.4 (CI 8.0-8.9) for persons with cancer in 2021. Compared with 2015-2019, MRRs for persons with mental disorders (anxiety, depression) and diabetes were significantly higher than expected in 2020. Interestingly, mortality for persons with mental disorders was significantly lower than expected in 2021. Significant mortality deficits were also observed for persons with cancer and chronic respiratory conditions (COPD, asthma) in 2020 and 2021. After controlling for social characteristics, only the mortality deficits for persons with cancer and chronic respiratory conditions remained significant (p < 0.005), with respectively, 17% and 11% lower mortality in 2020, and 12% and 6% lower mortality in 2021. Conclusions Large mortality differences were observed by pre-existing health problem and social background during the COVID-19 crisis. A significant mortality deficit was observed for persons with cancer and chronic respiratory conditions in 2020 and in 2021. Key messages • Information on pre-existing health problems and social ...