Smartphone applications (apps) for antibiotic prescribing have been studied in observational studies. Here, we assessed whether use of a smartphone app increased appropriateness of antibiotic therapy (AAT) in hospitals.An antibiotic stewardship app (Firstline.org) was populated with local guidelines and tested in a stepped-wedged cluster randomised trial in three hospitals in the Netherlands, Sweden and Switzerland. Defined clusters were randomised per hospital for the intervention (use of app) or standard of care. Primary outcome was AAT assessed by chart review in 15 random patients per cluster per intervention period. Secondary outcomes included clinical outcomes and user analytics. A questionnaire identifying barriers to app use was disseminated. Multivariable multilevel logistic models with time periods as fixed effects to adjust for time trend and treatment as fixed effects were employed to estimate the odds ratio of treatment.Twelve clusters in the Netherlands (1085 patients) were included, 12 in Sweden (362 patients) and 8 in Switzerland (653 patients). Overall, AAT was not increased (2.0% ([95%CI: -5.92 to 9.97]) in the intervention arm compared to control across the 3 centres. Average app use by cluster was associated with an AAT increase (1.9% [95%CI: 1.18 - 2.62]) across study centres; 3.2% in the Netherlands (p<0.01), 2.8% in Switzerland (p<0.01), and remained similar in Sweden (0.4%; p= 0.46). No difference was found for the other secondary outcomes. Main barriers for app use reported in the questionnaire were easily forgetting using the app and having other tools to help prescribing antibiotics.Overall, the introduction of a stewardship app did not significantly increase AAT, but a pre-specified secondary analysis of app use frequency was associated with a small but significant improvement of AAT. Variable uptake of the app, co-existing routes to guidelines and the impact of the COVID-19 pandemic during the trial likely had an impact on the results.ClinicalTrials.gov, trial number NCT03793946.JPIAMR, grant number JPIAMR2017-045.