Background: Neuraminidase inhibitors (NAIs) are considered the standard of care for hospitalised patients with influenza. We aimed to test whether combining the cap-dependent endonuclease inhibitor baloxavir marboxil (hereafter baloxavir) with standard-of-care NAIs would result in improved clinical outcomes compared with NAI monotherapy in hospitalised patients with severe influenza., Methods: We did a randomised, parallel-group, double-blind, placebo-controlled, superiority trial. Patients aged 12 years or older who were hospitalised with laboratory-confirmed influenza (by RT-PCR or a rapid test) and had a National Early Warning Score 2 (NEWS2) of 4 or greater were included. Recruitment took place in 124 centres across 25 countries. Using a permuted-block method and an interactive response system, patients were randomly assigned (2:1) to receive either baloxavir plus NAIs (hereafter the baloxavir group) or placebo plus NAIs (hereafter the control group). Participants, investigators, and those assessing outcomes were masked to group assignment. Baloxavir was administered orally on day 1 and day 4 (40 mg for bodyweight <80 kg, or 80 mg for ≥80 kg), and on day 7 if no clinical improvement had occurred by day 5. The NAIs included in this study were oseltamivir, zanamivir, and peramivir, which were selected and administered according to local standard practice. The primary endpoint was time to clinical improvement, defined as time to a NEWS2 of 2 or lower for 24 h or hospital discharge, whichever came first, based on daily assessments over the study duration of 35 days. Secondary endpoints included safety analyses. The modified intention-to-treat infected (mITTI) population (ie, all patients who were randomly assigned to treatment, received a dose of study drug, and were RT-PCR-positive for influenza at any timepoint according to the treatment assigned at randomisation) was used in all efficacy analyses. The safety population (ie, all patients who received at least one dose of study treatment, according to the treatment received) was used in the safety analyses. The trial is registered with ClinicalTrials.gov, NCT03684044., Findings: Overall, 366 patients were enrolled between Jan 8, 2019, and March 16, 2020, of whom 241 were assigned to the baloxavir group and 125 to the control group. The mITTI population comprised 322 patients, 208 in the baloxivir group and 114 in the control group. In total, 280 (87%) of these patients had influenza A infections. Median time to clinical improvement was 97·5 h (95% CI 75·9 to 117·2) in the baloxavir group and 100·2 h (75·9 to 144·4) in the control group (median difference -2·7 h [95% CI -53·4 to 25·9], p=0·467). Baloxavir plus NAI was well tolerated, and no new safety signals were observed; serious adverse events occurred in 29 (12%) of 239 patients in the baloxavir group versus 19 (15%) of 124 patients in the control group, of which one was considered related to treatment (orthostatic hypotension in a patient in the control group). Overall, four deaths (2%) occurred in the baloxavir group and seven (6%) in the control group; none were considered related to treatment., Interpretation: Combining baloxavir with NAIs did not result in superior clinical outcomes compared with NAIs alone. The combination of baloxavir plus NAI was well tolerated. The findings suggest that combination antivirals would not be routinely indicated in clinical practice for hospitalised patients with severe influenza., Funding: F Hoffmann-La Roche and the Biomedical Advanced Research and Development Authority., Competing Interests: Declaration of interests DK has received advisory fees from F Hoffmann-La Roche, Sanofi Pasteur, and GlaxoSmithKline, and a clinical trials grant from F Hoffmann-La Roche and GlaxoSmithKline. MGI received research support, paid to Northwestern University, from AiCuris, Janssen, and Shire; he is a paid consultant for Adagio, AlloVir, Celltrion, Cidara, Genentech, F Hoffmann-La Roche, Janssen, Shionogi, and Viracor Eurofins, and he is a paid member of data monitoring and safety boards for Janssen, Merck, SAB Biotherapeutics, Sequiris, Takeda, and Vitaeris. J-PM has received advisory fees from F Hoffmann-La Roche, AM-Pharma, and AKPA; lecture fees from Fresenius, MSD, Astellas, and Estor; and clinical trial grants from AKPA and Biomérieux. NL has previously received honoraria for consultancy work and speaking in educational programmes from Shionogi, Janssen, Sanofi Pasteur, Gilead Sciences Canada, F Hoffmann-La Roche, Genentech, and CIDARA Therapeutics, and travel support from Shionogi, Sanofi Pasteur, F Hoffmann-La Roche, and Genentech. LK, NC, AA, BC, and SWill are employees of Roche Products. SWild is an employee of F Hoffmann-La Roche. TW has received honoraria for consultancy work and speaking in educational programmes from AstraZeneca, Basilea, Biotest, Bayer, Boehringer Ingelheim, Berlin Chemie, GlaxoSmithKline, Infectopharm, Johnson & Johnson, MSD, Novartis, Pfizer, F Hoffmann-La Roche, and Sanofi Aventis. TS has received lecture fees from AstraZeneca, Boehringer Ingelheim, Sanofi Pasteur, Novartis, KYORIN Pharmaceutical, and Daiichi Sankyo. DSH, NZ, and JHH have nothing to declare., (Copyright © 2022 Elsevier Ltd. All rights reserved.)