A vaccine that prevents cytomegalovirus (CMV) infection in women could reduce the incidence of congenital CMV infection, a major cause of neurodevelopmental disability. We aimed to assess the safety and efficacy of a replication-defective investigational CMV vaccine, V160, in CMV-seronegative women. This phase 2b, randomised, double-blind, placebo-controlled study was conducted at 90 sites in seven countries (USA, Finland, Canada, Israel, Spain, Russia, and Australia). Eligible participants were generally healthy, CMV-seronegative, non-pregnant, 16-35-year-old women of childbearing potential with exposure to children aged 5 years or younger. Participants were randomly assigned using central randomisation via an interactive response technology system 1:1:1 to one of three groups: V160 three-dose regimen (V160 at day 1, month 2, and month 6), V160 two-dose regimen (V160 on day 1, placebo at month 2, and V160 at month 6), or placebo (saline solution at day 1, month 2, and month 6). The primary outcomes were the efficacy of three doses of V160 in reducing the incidence of primary CMV infection during the follow-up period starting 30 days after the last dose of vaccine using a fixed event rate design, and the safety and tolerability of the two-dose and three-dose V160 regimens. We planned to test the efficacy of a two-dose regimen of V160 in reducing the incidence of primary CMV infection only if the primary efficacy hypothesis was met. Analyses for the primary efficacy endpoint were performed on the per-protocol efficacy population; safety analyses included all randomly assigned participants who received study vaccine. The primary efficacy hypothesis was tested at prespecified interim and final analyses. The study was ongoing and efficacy data continued to accrue at the time of final testing of the primary efficacy hypothesis. Vaccine efficacy was re-estimated after final testing of the primary efficacy hypothesis based on all available efficacy data at end of study. This trial is registered at ClinicalTrials.gov (NCT03486834) and EudraCT (2017-004233-86) and is complete. Between April 30, 2018, and Aug 30, 2019, 7458 participants were screened, of whom 2220 were randomly assigned to the V160 three-dose group (n=733), V160 two-dose group (n=733), or placebo group (n=734). A total of 523 participants in the V160 three-dose group and 519 in the placebo group were included in the final hypothesis testing. Of these, there were 11 cases of CMV infection in the V160 three-dose group and 20 cases in the placebo group. The vaccine efficacy for the V160 three-dose group was 44·6% (95% CI -15·2 to 74·8) at the final testing of the primary efficacy hypothesis, a result corresponding to failure to demonstrate the primary efficacy hypothesis. On the basis of this result, the study was terminated for futility. The re-estimate of vaccine efficacy for the V160 three-dose group based on all available efficacy data at end of study (556 participants in the V160 three-dose group and 543 in the placebo group) was 42·4% (95% CI -13·5 to 71·1). A total of 728 participants in the V160 three-dose group, 729 in the V160 two-dose group, and 732 in the placebo group were included in the safety analyses. The most common solicited injection-site adverse event was injection-site pain (680 [93%] in the V160 three-dose group, 659 [90%] in the V160 two-dose group, and 232 [32%] in the placebo group). The most common solicited systemic adverse event was fatigue (457 [63%] in the V160 three-dose group, 461 [63%] in the V160 two-dose group, and 357 [49%] in the placebo group). No vaccine-related serious adverse events or deaths were reported. V160 was generally well tolerated and immunogenic; however, three doses of the vaccine did not reduce the incidence of primary CMV infection in CMV-seronegative women compared with placebo. This study provides insights into the design of future CMV vaccine efficacy trials, particularly for the identification of CMV infection using molecular assays. Merck Sharp & Dohme, a subsidiary of Merck & Co, Rahway, NJ, USA (MSD).