Background COVID-19 vaccines with alternative strain compositions are needed to provide broad protection against newly emergent SARS-CoV-2 variants of concern. This study aimed to describe the clinical efficacy and safety of a bivalent SARS-CoV-2 recombinant protein vaccine as a two-injection primary series during a period of circulation of the omicron (B.1.1.529) variant.Methods We conducted a phase 3, parallel, randomised, modified double-blind, placebo-controlled trial in adults aged 18 years or older at 54 clinical research centres in eight countries (Colombia, Ghana, India, Kenya, Mexico, Nepal, Uganda, and Ukraine). Participants were recruited from the community and randomly assigned (1:1) by use of an interactive response technology system to receive two intramuscular 05 mL injections, 21 days apart, of the bivalent vaccine (5 mu g of ancestral [D614] and 5 mu g of beta [B.1.351] variant spike protein, with AS03 adjuvant) or placebo (09% normal saline). All participants, outcome assessors, and laboratory staff performing assays were masked to group assignments; those involved in the preparation and administration of the vaccines were unmasked. Participants were stratified by age (18-59 years and >= 60 years) and baseline SARS-CoV-2 rapid serodiagnostic test positivity. Symptomatic COVID-19 was defined as laboratory-confirmed (via nucleic acid amplification test or PCR test) COVID-19 with COVID-19-like illness symptoms. The primary efficacy endpoint was the clinical efficacy of the bivalent vaccine for prevention of symptomatic COVID-19 at least 14 days after the second injection (dose 2). Safety was assessed in all participants receiving at least one injection of the study vaccine or placebo. This trial is registered with ClinicalTrials.gov (NCT04904549) and is closed to recruitment.Findings Between Oct 19, 2021, and Feb 15, 2022, 13 002 participants were enrolled and randomly assigned to receive the first dose of the study vaccine (n=6512) or placebo (n=6490). 12 924 participants (6472 in the vaccine group and 6452 in the placebo group) received at least one study injection, of whom 7542 (584%) were male and 9693 (750%) were SARS-CoV-2 non-naive. Of these 12 924 participants, 11 543 (893%) received both study injections (5788 in the vaccine group and 5755 in the placebo group). The efficacy-evaluable population after dose 2 comprised 11 416 participants (5736 in the vaccine group and 5680 in the placebo group). The median duration of follow-up was 85 days (IQR 50-95) after dose 1 and 58 days (29-70) after dose 2. 121 symptomatic COVID-19 cases were reported at least 14 days after dose 2 (32 in the vaccine group and 89 in the placebo group), with an overall vaccine efficacy of 647% (95% CI 466 to 772). Vaccine efficacy against symptomatic COVID-19 was 751% (95% CI 563 to 866) in SARS-CoV-2 non-naive participants and 309% (-393 to 667) in SARS-CoV-2-naive participants. Viral genome sequencing identified the infecting strain in 68 (562%) of 121 cases (omicron [BA.1 and BA.2] in 63; delta in four; and both omicron and delta in one). Immediate unsolicited adverse events were reported by four (<01%) participants in the vaccine group and seven (01%) participants in the placebo group. Immediate unsolicited adverse reactions within 30 min after any injection were reported by four (<01%) participants in the vaccine group and six (<01%) participants in the placebo group. In the reactogenicity subset with available data, solicited reactions (solicited injection-site reactions and solicited systemic reactions) within 7 days after any injection occurred in 1398 (578%) of 2420 vaccine recipients and 983 (409%) of 2403 placebo recipients. Grade 3 solicited reactions were reported by 196 (81%; 95% CI 70 to 93) of 2420 vaccine recipients and 118 (49%; 41 to 59) of 2403 placebo recipients within 7 days after any injection, with comparable frequencies after dose 1 and dose 2 in the vaccine group. At least one serious adverse event occurred in 30 (05%) participants in the vaccine group and 26 (04%) in the placebo group. The proportion of adverse events of special interest and deaths was less than 01% in both study groups. No adverse event of special interest, serious adverse event, or death was deemed to be treatment related. There were no reported cases of thrombosis with thrombocytopenia syndrome, myocarditis, pericarditis, Bell's Palsy, or Guillain-Barr & eacute; syndrome, or other immune-mediated diseases.Interpretation The bivalent variant vaccine conferred heterologous protection against symptomatic SARS-CoV-2 infection in the epidemiological context of the circulating contemporary omicron variant. These findings suggest that vaccines developed with an antigen from a non-predominant strain could confer cross-protection against newly emergent SARS-CoV-2 variants, although further investigation is warranted.