There are no approved pharmacological therapies to support treatment of the core communication and socialisation difficulties associated with autism spectrum disorder in adults. We aimed to assess the efficacy, safety, and pharmacokinetics of balovaptan, a vasopressin 1a receptor antagonist, versus placebo in autistic adults. V1aduct was a phase 3, randomised, placebo-controlled, double-blind trial, conducted at 46 sites across six countries (the USA, the UK, France, Italy, Spain, and Canada). Eligible participants were aged 18 years or older with an intelligence quotient (IQ) of 70 or higher, and met the criteria for moderate-to-severe autism spectrum disorder (DSM-5 and Autism Diagnostic Observation Schedule). Participants were randomly allocated (1:1), with an independent interactive voice or web-based response system, to receive balovaptan (10 mg) or placebo daily for 24 weeks. Randomisation was stratified by an individual's baseline Vineland-II two-domain composite (2DC) score (<60 or ≥60), sex, region (North America or rest of world), and age (<25 years or ≥25 years). Participants, study site personnel, and the sponsor were masked to treatment assignment. The primary endpoint was change from baseline in Vineland-II 2DC score (the mean composite score across the Vineland-II socialisation and communication domains) at week 24. The primary analysis was done with ANCOVA in the intention-to-treat population. The V1aduct study was terminated for futility after around 50% of participants completed the week 24 visit. This trial is registered with ClinicalTrials.gov (NCT03504917). Between Aug 8, 2018, and July 1, 2020, 540 people were screened for eligibility, of whom 322 were allocated to receive balovaptan (164 [51%]) or placebo (158 [49%]). One participant from the balovaptan group was not treated before trial termination and was excluded from the analysis. 60 participants in the balovaptan group and 55 in the placebo group discontinued treatment before week 24. The sample consisted of 64 (20%) women and 257 (80%) men, with 260 (81%) participants from North America and 61 (19%) from Europe. At baseline, mean age was 27·6 years (SD 9·7) and mean IQ score was 104·8 (18·1). Two (1%) participants were American Indian or Alaska Native, eight (2%) were Asian, 15 (5%) were Black or African American, 283 (88%) were White, four (1%) were of multiple races, and nine (3%) were of unknown race. Mean baseline Vineland-II 2DC scores were 67·2 (SD 15·3) in the balovaptan group and 66·2 (17·7) in the placebo group. The interim futility analysis showed no improvement for balovaptan versus placebo in terms of Vineland-II 2DC score at week 24 compared with baseline, with a least-squares mean change of 2·91 (SE 1·52) in the balovaptan group (n=79) and 4·75 (1·60) in the placebo group (n=71; estimated treatment difference -1·84 [95% CI -5·15 to 1·48]). In the final analysis, mean change from baseline in Vineland-II 2DC score at week 24 was 4·56 (SD 10·85) in the balovaptan group (n=111) and 6·83 (12·18) in the placebo group (n=99). Balovaptan was well tolerated, with similar proportions of participants with at least one adverse event in the balovaptan group (98 [60%] of 163) and placebo group (104 [66%] of 158). The most common adverse events were nasopharyngitis (14 [9%] in the balovaptan group and 19 [12%] in the placebo group), diarrhoea (11 [7%] and 14 [9%]), upper respiratory tract infection (ten [6%] and nine [6%]), insomnia (five [3%] and eight [5%]), oropharyngeal pain (five [3%] and eight [5%]), and dizziness (two [1%] and ten [6%]). Serious adverse events were reported for two (1%) participants in the balovaptan group (one each of suicidal ideation and schizoaffective disorder), and five (3%) participants in the placebo group (one each of suicidal ideation, panic disorder, limb abscess, urosepsis, colitis [in the same participant with urosepsis], and death by suicide). No treatment-related deaths occurred. Balovaptan did not improve social communication in autistic adults. This study provides insights into challenges facing autism spectrum disorder trials, including the considerable placebo response and the selection of appropriate outcome measures. F Hoffmann-La Roche.