Abstract Background Emerging evidence shows the substantial real-world impact of authorised vaccines against COVID-19 and provides insight into the potential role of vaccines in curbing the pandemic. However, there remains uncertainty about the efficacy of vaccines against different variants of the virus. Here we assessed efficacy of ChAdOx1 nCoV-19 (AZD1222) against lineages of SARS-CoV-2 circulating in Brazil from June 2020 until early 2021. Methods Participants aged 18 and above were enrolled into a randomised phase 3 trial of ChAdOx1 nCoV-19 vaccine against symptomatic SARS-CoV-2 infection. Participants received two doses of ChAdOx1 nCoV-19 or control (1st dose: Men ACWY vaccine, 2nd dose: normal saline). Nasopharyngeal and oropharyngeal swabbing was performed if participants developed symptoms of COVID-19 (cough, shortness of breath, fever >37.8°C, ageusia, anosmia). Swabs were tested by nucleic acid amplification (NAAT) for SARS-CoV-2, sequenced, and viral load determined. For those samples where a genotype could not be ascertained from sequencing, allele specific PCR was performed. The efficacy analysis included symptomatic COVID-19 in seronegative participants with a NAAT positive swab more than 14 days after a second dose of vaccine. Participants were unblinded after the vaccine was authorised for use, and the control participants offered vaccination. Infections occurring after unblinding were excluded from analysis. Vaccine efficacy was calculated as 100% x (1 – relative risk (RR)), where RR was estimated from a robust Poisson model. The trial is registered at ISRCTN89951424. Findings 9433 participants were eligible for inclusion in the pre-specified primary efficacy population, having reached more than 14 days after a second dose of ChAdOx1 nCoV-19, of whom 307 were NAAT+, in this post-hoc analysis. From June 2020 to February 2021, the two most frequently identified lineages were P.2 (N=153) and B.1.1.28 (N=49). P.1 emerged during the study (N=18) but became dominant only after study unblinding. Viral loads were highest amongst those with P.1 infection. Vaccine efficacy (VE) for B.1.1.33 (88.2%, 95%CI 5, 99), B.1.1.28 (73%, 95% CI, 46, 86), P.2 (69% 95% CI, 55, 78) and P.1 (64%, 95% CI, -2, 87) was estimated. In participants who had received two doses of vaccine, one COVID-19 hospitalisation occurred in the ChAdOx1 nCoV-19 group and 18 in the control group, with VE against hospitalisation 95% (95% CI 61, 99). There were 2 COVID-19 deaths in the control group and none in the vaccine group. Interpretation ChAdOx1 nCoV-19 provides high efficacy against hospitalisation, severe disease and death from COVID-19 in Brazil and there is strong evidence of protection being maintained against P.2, despite the presence of the spike protein mutation E484K. Real world effectiveness studies are ongoing in Brazil to further establish protection against P.1 and other emerging variants.