Amisulpride, aripiprazole, and olanzapine are first-line atypical antipsychotics that have not previously been compared head-to-head in a pragmatic trial. We aimed to compare the efficacy and safety of these agents in a controlled trial.This pragmatic, rater-blind, randomised controlled trial was done in three academic centres of psychiatry in Norway, and one in Austria. Eligible patients were aged 18 years or older, met ICD-10 criteria for schizophrenia-spectrum disorders (F20-29), and had symptoms of active psychosis. Eligible patients were randomly assigned to receive oral amisulpride, aripiprazole, or olanzapine. Treatment allocation was open to patients and staff, and starting dose, treatment changes, and adjustments were left to the discretion of the treating physician. Computer-generated randomisation lists for each study centre were prepared by independent statisticians. Patients were followed up for 52 weeks after random assignment, during which assessments were done 8 times by researchers masked to treatment. The primary outcome was reduction of the Positive And Negative Syndrome Scale (PANSS) total score at 52 weeks, and primary analyses were done in the intention-to-treat population. This trial is registered with ClinicalTrials.gov, number NCT01446328.Between Oct 20, 2011, and Dec 30, 2016, we assessed 359 patients for eligibility. 215 patients were excluded (107 did not meet inclusion criteria, 82 declined to participate, 26 other reasons). 144 patients (mean baseline PANSS total estimated score 78·4 [SD 1·4]) were randomly assigned 1:1:1 to receive amisulpride (44 patients), aripiprazole (48 patients) or olanzapine (52 patients). After 52 weeks, the patients allocated to amisulpride had a PANSS total score reduction of 32·7 points (SD 3·1) compared with 21·9 points reduction with aripiprazole (SD 3·9, p=0·027) and 23·3 points with olanzapine (2·9, p=0·025). We observed weight gain and increases of serum lipids and prolactin in all groups. 26 serious adverse events (SAEs) among 20 patients were registered (four [9%] of 44 patients allocated to amisulpride, ten [21%] of 48 patients allocated to aripiprazole, and six [12%] of 52 patients allocated to olanzapine), with no statistically significant differences between the study drugs. 17 (65%) of the 26 SAEs occurred during the use of the study drug, with readmission or protracted hospital admission accounting for 13 SAEs. One death by suicide, one unspecified death, and one life-threatening accident occurred during follow-up, after cessation of treatment.Amisulpride was more efficacious than aripiprazole or olanzapine for reducing the PANSS total scores in adults with schizophrenia-spectrum disorders. Side-effect differences among the groups were generally small. This study supports the notion that clinically relevant efficacy differences exist between antipsychotic drugs. Future research should aim to compare first-line antipsychotics directly in pragmatic clinical trials that reflect everyday clinical practice.The Research Council of Norway, the Western Norway Regional Health Trust, and participating hospitals and universities.