BackgroundUser-testing and subsequent modification of clinical guidelines increases health professionals’ information retrieval and comprehension. No study has investigated whether this results in safer care.ObjectiveTo compare the frequency of medication errors when administering an intravenous medicine using the current National Health Service Injectable Medicines Guide (IMG) versus an IMG version revised with user-testing.MethodSingle-blind, randomised parallel group in situ simulation. Participants were on-duty nurses/midwives who regularly prepared intravenous medicines. Using a training manikin in their clinical area, participants administered a voriconazole infusion, a high-risk medicine requiring several steps to prepare. They were randomised to use current IMG guidelines or IMG guidelines revised with user-testing. Direct observation was used to time the simulation and identify errors. Participant confidence was measured using a validated instrument. The primary outcome was the percentage of simulations with at least one moderate-severe IMG-related error, with error severity classified by an expert panel.ResultsIn total, 133 participants were randomised to current guidelines and 140 to user-tested guidelines. Fewer moderate-severe IMG-related errors occurred with the user-tested guidelines (n=68, 49%) compared with current guidelines (n=79, 59%), but this difference was not statistically significant (risk ratio: 0.82; 95% CI 0.66 to 1.02). Significantly more simulations were completed without any IMG-related errors with the user-tested guidelines (n=67, 48%) compared with current guidelines (n=26, 20%) (risk ratio: 2.46; 95% CI 1.68 to 3.60). Median simulation completion time was 1.6 min (95% CI 0.2 to 3.0) less with the user-tested guidelines. Participants who used user-tested guidelines reported greater confidence.ConclusionUser-testing injectable medicines guidelines reduces the number of errors and the time taken to prepare and administer intravenous medicines, while increasing staff confidence.Trial registration numberresearchregistry5275.