BACKGROUND: Although intravenous immunoglobulin (IVIG) is effective therapy for Kawasaki disease, 10-20% of patients have recrudescent fever as a sign of persistent inflammation and require additional treatment. We aimed to compare infliximab with a second infusion of IVIG for treatment of resistant Kawasaki disease. METHODS: In this multicentre comparative effectiveness trial, patients (aged 4 weeks to 17 years) with IVIG resistant Kawasaki disease and fever at least 36 h after completion of their first IVIG infusion were recruited from 30 hospitals across the USA. Patients were randomly assigned (1:1) to second IVIG (2 g/kg over 8-12 h) or intravenous infliximab (10 mg/kg over 2 h without premedication), by using a randomly permuted block randomisation design with block size of two or four. Patients with fever 24 h to 7 days following completion of first study treatment crossed over to receive the other study treatment. The primary outcome measure was resolution of fever at 24 h after initiation of study treatment with no recurrence of fever attributed to Kawasaki disease within 7 days post-discharge. Secondary outcome measures included duration of fever from enrolment, duration of hospitalisation after randomisation, and changes in markers of inflammation and coronary artery Z score. Efficacy was analysed in participants who received treatment and had available outcome values. Safety was analysed in all randomised patients who did not withdraw consent. This clinical trial is registered with ClinicalTrials.gov, NCT03065244. FINDINGS: Between March 1, 2017, and Aug 31, 2020, 105 patients were randomly assigned to treatment and 103 were included in the intention-to-treat population (54 in the infliximab group, 49 in the second IVIG group). Two patients randomised to infliximab did not receive allocated treatment. The primary outcome was met by 40 (77%) of 52 patients in the infliximab group and 25 (51%) of 49 patients in the second IVIG infusion group (odds ratio 0·31, 95% CI 0·13-0·73, p=0·0076). 31 patients with fever beyond 24 h received crossover treatment: nine (17%) in the infliximab group received second IVIG and 22 (45%) in second IVIG group received infliximab (p=0·0024). Three patients randomly assigned to infliximab and two to second IVIG with fever beyond 24h did not receive crossover treatment. Mean fever days from enrolment was 1·5 (SD 1·4) for the infliximab group and 2·5 (2·5) for the second IVIG group (p=0·014). Mean hospital stay was 3·2 days (2·1) for the infliximab group and 4·5 days (2·5) for the second IVIG group (p<0·001). There was no difference between treatment groups for markers of inflammation or coronary artery outcome. 24 (44%) of 54 patients in the infliximab group and 33 (67%) of 49 in the second IVIG group had at least one adverse event. A drop in haemoglobin concentration of at least 2g/dL was seen in 19 (33%) of 58 patients who received IVIG as either their first or second study treatment (three of whom required transfusion) and in three (7%) of 43 who received only infliximab (none required transfusion; p=0·0028). Haemolytic anaemia was the only serious adverse events deemed definitely or probably related to study treatment, and was reported in nine (15%) of 58 patients who received IVIG as either their first or second study treatment and none who received infliximab only. INTERPRETATION: Infliximab is a safe, well tolerated, and effective treatment for patients with IVIG resistant Kawasaki disease, and results in shorter duration of fever, reduced need for additional therapy, less severe anaemia, and shorter hospitalisation compared with second IVIG infusion. FUNDING: Patient Centered Outcomes Research Institute.