Abstract Background In Serbia, premature mortality due to COVID-19 usually has been assessed using aggregated mortality data, thus overlooking the impact of sociodemographic factors. Within the European Burden of Disease Network (COST Action CA18218), we examined the association between several sociodemographic factors and COVID-19 years of life lost (YLL) to reveal potential social inequalities. Methods The Statistical Office of the Republic of Serbia (SORS) has provided disaggregated, anonymized, and individual data on COVID-19 deaths in 2020 and 2021 from death certificates having U07.1 and U07.2 ICD codes as the main cause of death. Additionally, SORS remaining life expectancy and population size estimates were used along with sociodemographic data for each YLL case (sex, time and place of death, type of residential settlement, educational attainment, marital status, and economic activity). Results In the observed two-year period, 38,098 COVID-19 death cases were assessed, of which 72.8% were registered in 2021. The average YLL per death case was 12.51±7.62, and the median YLL was 11.50 (12.60±7.44 for men and 12.39±7.85 for women, p < 0.006). The multivariate linear regression indicated a significant association between the number of YLL and sex (B = 1.02, p < 0.01), residential area (B = 0.21, p < 0.01), educational attainment (B = 0.84, p < 0.01), marital status (B = 1.65, p < 0.01), employment (B=-7.02, p < 0.01) and place of death (B=-1.76, p < 0.01). Conclusions This study demonstrates social inequalities in COVID-19 premature mortality associated with the highest YLL among men, single persons, retired persons, and those with higher education status. We identify significant differences in YLL when comparing different regions. Key messages • Individual-level data reveals differences in premature mortality aggravated by social disparities. • Policymakers should consider sociodemographic factors when designing COVID-19 prevention measures.