Anti-tumour necrosis factor (TNF) agents are the mainstay of long-term treatment for refractory ulcerative colitis. However, long-term use of anti-TNF therapy might lead to an increased risk of malignancy or infection. To date, no randomised controlled trial has evaluated whether anti-TNF agents can be safely discontinued in patients with ulcerative colitis in remission. We therefore aimed to compare outcomes in these patients who continued infliximab with those who discontinued infliximab. We did a multicentre, open-label randomised controlled trial at 24 specialist centres in Japan. We enrolled patients with ulcerative colitis who were in remission, had been treated with intravenous infliximab (5 mg/kg) every 8 weeks, and had started infliximab at least 14 weeks before study enrolment. No restrictions regarding age and comorbidities were used to exclude participation. Patients who were confirmed to be in remission for more than 6 months, to be corticosteroid-free, and to have a Mayo Endoscopic Subscore (MES) of 0 or 1 were centrally randomised. An independent organisation randomly assigned patients (1:1) into either the infliximab-continued group or infliximab-discontinued group, using a computer-generated stratified randomisation procedure. The stratified factors were the use of immunomodulators (yes or no) and MES (0 or 1). Neither patients nor health-care providers were masked to the randomisation. The primary endpoint was the remission rate at week 48 in the full analysis set, which was based on the intention-to-treat principle and excluded participants with no efficacy data after randomisation. This study was registered with the University Hospital Medical Information Network Center Trials registry, UMIN000012092. Between June 16, 2014, and July 28, 2017, 122 patients were eligible for screening and a total of 95 patients were randomly assigned to the infliximab-continued group (n=48) or the infliximab-discontinued group (n=47). 92 patients (n=46 for both groups) were included in the full analysis set. 37 (80·4% [95% CI 66·1-90·6]) of 46 patients in the infliximab-continued group and 25 (54·3% [39·0-69·1]) of 46 patients in the infliximab-discontinued group were in remission at week 48. The between-group difference was 26·1% (95% CI 7·7-44·5; p=0·0076) before adjustment and 27·3% (95% CI 8·0-44·1; p=0·0059) after adjustment for stratification factors. Eight (17%) of 48 patients in the infliximab-continued group and six (13%) of 47 in the infliximab-discontinued group developed adverse events (between-group difference 3·9% [95% CI -10·3 to 18·1]; p=0·59). In the infliximab-continued group, one patient had an infusion reaction and two patients had psoriatic skin lesions. Eight (66·7%, 95% CI 34·9-90·1) of the 12 patients in the infliximab-discontinuation group who were re-treated with infliximab after relapsing were in remission within 8 weeks of re-treatment; none had infusion reactions. Maintenance of remission was significantly more common in patients who continued infliximab than in those who discontinued. Discontinuing infliximab should therefore be discussed with caution, taking both risk of relapse and efficacy of re-treatment into account. Mitsubishi Tanabe Pharma Corporation and the Intractable Disease Project of the Ministry of Health, Labour and Welfare of Japan. For the Japanese translation of the abstract see Supplementary Materials section.