Inclusion of additional influenza A/H3N2 strains in seasonal influenza vaccines could expand coverage against multiple, antigenically distinct, cocirculating A/H3N2 clades and potentially replace the no longer circulating B/Yamagata strain. We aimed to evaluate the safety and immunogenicity of three next-generation seasonal influenza mRNA vaccines with different compositions that encode for haemagglutinins of multiple A/H3N2 strains, with or without the B/Yamagata strain, in adults. This randomised, open-label, phase 1/2 trial enrolled healthy adults aged 50-75 years across 22 sites in the USA. Participants were randomly assigned (1:1:1:1:1:1:1) via interactive response technology to receive a single dose of mRNA-1011.1 (pentavalent; containing one additional A/H3N2 strain [Newcastle]), mRNA-1011.2 (quadrivalent; B/Yamagata replaced with one additional A/H3N2 strain [Newcastle]), mRNA-1012 at one of two dose levels (pentavalent; B/Yamagata replaced with two additional A/H3N2 strains [Newcastle and Hong Kong]), or one of three quadrivalent mRNA-1010 controls each encoding one of the A/H3N2 study strains. The primary outcomes were safety, evaluated in all randomly assigned participants who received a study vaccination (safety population), and reactogenicity, evaluated in all participants from the safety population who contributed any solicited adverse reaction data (solicited safety population). The secondary outcome was humoral immunogenicity of investigational mRNA vaccines at day 29 versus mRNA-1010 control vaccines based on haemagglutination inhibition antibody (HAI) assay in the per-protocol population. Here, we summarise findings from the planned interim analysis after participants had completed day 29. The study is registered with ClinicalTrials.gov, NCT05827068, and is ongoing. Between March 27 and May 9, 2023, 1183 participants were screened for eligibility, 699 (59·1%) were randomly assigned, and 696 (58·8%) received vaccination (safety population, n=696; solicited safety population, n=694; per-protocol population, n=646). 382 (55%) of the 696 participants in the safety population self-reported as female and 314 (45%) as male. Frequencies of solicited adverse reactions were similar across vaccine groups; 551 (79%) of 694 participants reported at least one solicited adverse reaction within 7 days after vaccination and 83 (12%) of 696 participants reported at least one unsolicited adverse event within 28 days after vaccination. No vaccine-related serious adverse events or deaths were reported. All three next-generation influenza vaccines elicited robust antibody responses against vaccine-matched influenza A and B strains at day 29 that were generally similar to mRNA-1010 controls, and higher responses against additional A/H3N2 strains that were not included within respective mRNA-1010 controls. Day 29 geometric mean fold rises in HAI titres from day 1 against vaccine-matched A/H3N2 strains were 3·0 (95% CI 2·6-3·6; Darwin) and 3·1 (2·6-3·8; Newcastle) for mRNA-1011.1; 3·3 (2·7-4·1; Darwin) and 4·2 (3·4-5·2; Newcastle) for mRNA-1011.2; 3·4 (2·9-4·0; Darwin), 4·5 (3·6-5·5; Newcastle), and 5·1 (4·2-6·2; Hong Kong) for mRNA-1012 50·0 μg; and 2·6 (2·2-3·1; Darwin), 3·7 (3·0-4·6; Newcastle), and 4·1 (3·3-5·1; Hong Kong) for mRNA-1012 62·5 μg. Inclusion of additional A/H3N2 strains did not reduce responses against influenza A/H1N1 or influenza B strains, and removal of B/Yamagata did not affect responses to B/Victoria. These data support the continued clinical development of mRNA-based next-generation seasonal influenza vaccines with broadened influenza A/H3N2 strain coverage. Moderna.