ABSTRACT Background GBP510 vaccine contains self-assembling, recombinant nanoparticles displaying SARS-CoV-2 spike receptor-binding domains. We report interim phase 3 immunogenicity results for GBP510 adjuvanted with AS03 (GBP510/AS03) compared with ChAdOx1-S (Vaxzevria, AstraZeneca) up to 2 weeks after the second dose, and safety data up to a median of 2.5 months follow-up. Methods Randomised, active-controlled, observer-blinded, multinational study: Cohort 1 (no history of SARS-CoV-2 infection/COVID-19 vaccination, n=1956) randomised 2:1 to receive two doses of GBP510/AS03 or ChAdOx1-S (immunogenicity and safety); Cohort 2 (regardless of baseline serostatus; n=2080) randomised 5:1 (safety). Primary objectives: demonstrate superiority in geometric mean titre (GMT) and non-inferiority in seroconversion rate (SCR; ≥4-fold rise from baseline) of GBP510/AS03 versus ChAdOx1-S for neutralising antibodies against the ancestral strain by live-virus neutralisation assay. Secondary objectives included assessment of safety and reactogenicity. Findings At 2 weeks after the second vaccination, the GMT ratio (GBP510/AS03 / ChAdOx1-S) was 2.93 (95% CI 2.63–3.27), demonstrating superiority (95% CI lower limit >1). The between-group SCR difference of 10.76% (95% CI 7.68–14.32) satisfied the non-inferiority criterion (95% CI lower limit > −5%).The proportion of subjects with adverse events (AEs) after any vaccination was higher with GBP510/AS03 versus ChAdOx1-S for solicited local AEs (56.69% vs 49.20%), but was similar for solicited systemic AEs (51.21% vs 53.51%) and unsolicited AEs (13.27% vs 14.56%). No safety concerns were identified during follow-up for a median 2.5-months after the second vaccination. Interpretation GBP510/AS03 met the superiority criterion for neutralising antibodies and non-inferiority criterion for SCR compared with ChAdOx1-S, and showed a clinically acceptable safety profile. Funding This work was supported, in whole or in part, by funding from CEPI and the Bill & Melinda Gates Foundation Investments INV-010680 and INV-006462. The Bill & Melinda Gates Foundation supported this project for the generation of IND-enabling data and CEPI supported this clinical study. RESEARCH IN CONTEXT Evidence before this study Immunobridging has been proposed as an approach for assessing new COVID-19 vaccines by comparing the immunogenicity of candidate vaccines with an active comparator with demonstrated clinical efficacy. We searched PubMed up to 26 October 2022 for immunobridging clinical trials comparing a candidate vaccine with an approved vaccine, using the terms “immunobridging”, “SARS-CoV-2”, “COVID-19”, and “vaccine”. A phase 2/3 study showed that the ChAdOx1 vaccine, manufactured by the Serum Institute of India after technology transfer from Oxford University/AstraZeneca, had a non-inferior immune response compared to the original ChAdOx1 (AZD1222) vaccine. A post hoc analysis of phase 2 data found that MVC-COV1901 vaccine (a protein subunit vaccine developed by Medigen Vaccine Biologics Corporation, Taiwan) was non-inferior to ChAdOx1 (AZD1222) with respect to neutralising antibody titres. A phase 3 study found that VLA2001 (an adjuvanted, inactivated whole-virus vaccine developed by Valneva, Austria) was superior to ChAdOx1 with respect to neutralising antibody titres and non-inferior with respect to seroconversion rates. Added value of this study This is the first study comparing the immunogenicity of recombinant SARS-CoV-2 protein nanoparticle vaccine GBP510 adjuvanted with AS03 versus ChAdOx1-S. Interim analysis found that two-dose vaccination with GBP510/AS03 induced stronger neutralising antibody immune responses compared with ChAdOx1-S against the ancestral D614G strain at 2 weeks after the second dose. GBP510/AS03 had an acceptable safety profile during a median 2.5 months of follow-up. Implications of the available evidence This interim analysis suggests that GBP510/AS03 induces strong neutralising antibody responses against SARS-CoV-2 ancestral strain and has an acceptable safety profile.