BACKGROUND: Dual blockade of PD-L1 and VEGF has enhanced anticancer immunity through multiple mechanisms and augmented antitumour activity in multiple malignancies. We aimed to assess the efficacy and safety of atezolizumab (anti-PD-L1) alone and combined with bevacizumab (anti-VEGF) in patients with unresectable hepatocellular carcinoma. METHODS: GO30140 is an open-label, multicentre, multiarm, phase 1b study that enrolled patients at 26 academic centres and community oncology practices in seven countries worldwide. The study included five cohorts, and the two hepatocellular carcinoma cohorts, groups A and F, are described here. Inclusion criteria for these two groups included age 18 years and older; histologically, cytologically, or clinically (per American Association for the Study of Liver Diseases criteria) confirmed unresectable hepatocellular carcinoma that was not amenable to curative treatment; no previous systemic treatment; and Eastern Cooperative Oncology Group performance status of 0 or 1. In group A, all patients received atezolizumab (1200 mg) and bevacizumab (15 mg/kg) intravenously every 3 weeks. In group F, patients were randomly assigned (1:1) to receive intravenous atezolizumab (1200 mg) plus intravenous bevacizumab (15 mg/kg) every 3 weeks or atezolizumab alone by interactive voice-web response system using permuted block randomisation (block size of two) and stratification factors of geographical region; macrovascular invasion, extrahepatic spread, or both; and baseline α-fetoprotein concentration. Primary endpoints were confirmed objective response rate in all patients who received the combination treatment for group A and progression-free survival in the intention-to-treat population in group F, both assessed by an independent review facility according to Response Evaluation Criteria in Solid Tumors version 1.1. In both groups, safety was assessed in all patients who received at least one dose of any study treatment. This study is registered with ClinicalTrials.gov, NCT02715531, and is closed to enrolment. FINDINGS: In group A, 104 patients were enrolled between July 20, 2016, and July 31, 2018, and received atezolizumab plus bevacizumab. With a median follow-up of 12·4 months (IQR 8·0-16·2), 37 (36%; 95% CI 26-46) of 104 patients had a confirmed objective response. The most common grade 3-4 treatment-related adverse events were hypertension (13 [13%]) and proteinuria (seven [7%]). Treatment-related serious adverse events occurred in 25 (24%) patients and treatment-related deaths in three (3%) patients (abnormal hepatic function, hepatic cirrhosis, and pneumonitis). In group F, 119 patients were enrolled and randomly assigned (60 to atezolizumab plus bevacizumab; 59 to atezolizumab monotherapy) between May 18, 2018, and March 7, 2019. With a median follow-up of 6·6 months (IQR 5·5-8·5) for the atezolizumab plus bevacizumab group and 6·7 months (4·2-8·2) for the atezolizumab monotherapy group, median progression-free survival was 5·6 months (95% CI 3·6-7·4) versus 3·4 months (1·9-5·2; hazard ratio 0·55; 80% CI 0·40-0·74; p=0·011). The most common grade 3-4 treatment-related adverse events in group F were hypertension (in three [5%] patients in the atezolizumab plus bevacizumab group; none in the atezolizumab monotherapy group) and proteinuria (in two [3%] patients in the atezolizumab plus bevacizumab group; none in the atezolizumab monotherapy group). Treatment-related serious adverse events occurred in seven (12%) patients in the atezolizumab plus bevacizumab group and two (3%) patients in the atezolizumab monotherapy group. There were no treatment-related deaths. INTERPRETATION: Our study shows longer progression-free survival with a combination of atezolizumab plus bevacizumab than with atezolizumab alone in patients with unresectable hepatocellular carcinoma not previously treated with systemic therapy. Therefore, atezolizumab plus bevacizumab might become a promising treatment option for these patients. This combination is being compared with standard-of-care sorafenib in a phase 3 trial. FUNDING: F Hoffmann-La Roche/Genentech.