Objective: To estimate the impact of sociodemographic factors on survival from COVID-19 in Brazil. Methods: Longitudinal data from a retrospective cohort of 2,000,000 hospitalizations due to COVID-19 in Brazil between March 2020 and May 2022, enrolled in SIVEP-Gripe, were analyzed. Results: The adjusted Cox model showed a 7% higher probability of death for men. 9% and 13% for the brown population compared to white and 16% for those living in the rural region. Long-lived elderly has a 301% higher probability when compared to young people. Conclusion: Sociodemographic factors interfere with survival from COVID-19 and should gain prominence in theoretical models and clinical aspects, and should be considered when formulating public policies, especially in countries with greater social inequality, such as Brazil.