Emergence of the SARS-CoV-2 omicron (B.1.1.529) variant with high immune evasion has led to the development and roll-out of bivalent mRNA vaccines targeting original and omicron strains. However, real-world observational data on effectiveness of bivalent vaccines are scarce. We aimed to assess the relative effectiveness of a fourth vaccine dose with the BA.1-adapted or BA.4/BA.5-adapted bivalent vaccines against medically attended symptomatic SARS-CoV-2 infection and COVID-19-related hospital admission among SARS-CoV-2-naive and previously infected individuals in Singapore. We conducted a retrospective cohort study among Singapore residents aged 18 years and older who had received three monovalent mRNA vaccine doses and were eligible for a fourth dose. Data were collected from official databases on COVID-19 cases and vaccinations maintained by the Singapore Ministry of Health. We analysed the incidence of medically attended symptomatic SARS-CoV-2 infection and COVID-19-related hospital admission between Oct 14, 2022, and Jan 31, 2023, by previous infection status and type of fourth vaccine dose received. Inverse probability-weighted Cox regressions were used to estimate hazard ratios (HRs). 2 749 819 individuals were included in the analysis. For the SARS-CoV-2-naive group, a fourth monovalent vaccine dose did not confer additional protection over three monovalent doses against symptomatic infection (HR 1·09 [95% CI 1·07-1·11]), whereas the bivalent vaccine did provide additional protection (0·18 [0·17-0·19]). Among individuals with previous infection, the HR was 0·87 (95% CI 0·84-0·91) and 0·14 (0·13-0·15) with receipt of the fourth monovalent and bivalent doses, respectively. Against COVID-19-related hospital admission, the bivalent vaccine (HR 0·12 [95% CI 0·08-0·18] in SARS-CoV-2-naive participants and 0·04 [0·01-0·15] in previously infected participants) conferred greater benefit compared with the fourth monovalent dose (0·84 [0·77-0·91] in SARS-CoV-2-naive participants and 0·85 [0·69-1·04] in previously infected participants). A fourth dose with the bivalent vaccine was substantially more effective against medically attended symptomatic SARS-CoV-2 infection and COVID-19-related hospital admission than four monovalent doses among both SARS-CoV-2-naive and previously infected individuals. Boosters with the bivalent vaccine might be preferred in this omicron-predominant pandemic, regardless of previous infection history. None.