We assessed the safety of tacrolimus therapy for rheumatoid arthritis. Forty-two patients who started tacrolimus therapy between April 2005 and July 2006 were investigated retrospectively using data from their medical records up to June 2007. The cumulative treatment continuation rate was assessed by the Kaplan-Meier method. Fisher's exact test was used to compare gastrointestinal symptoms between different tacrolimus doses and between the presence and absence of each concomitant medication. The mean (+/-SD) observation period was 288 +/- 238 days. The cumulative treatment continuation rate was, respectively, 59.5% and 38.1% at 6 months and 1 year after the patients started treatment. Tacrolimus was discontinued in 28 patients, and was discontinued because of adverse reactions in 21 patients. Gastrointestinal symptoms were the most common adverse reactions (45.2% = 19/42 patients), followed by infections and hyperglycemia. Tacrolimus was discontinued in 9/19 patients with gastrointestinal symptoms, and was discontinued within 60 days of starting treatment in seven of them. Nausea and vomiting led to discontinuation in seven patients (within 60 days of starting treatment in six of them). The incidence of gastrointestinal symptoms was higher in patients receiving a daily dose >or=2 mg than in those receiving <2 mg/day. During treatment of rheumatoid arthritis by oral tacrolimus therapy, gastrointestinal symptoms were common, early, and dose-dependent. However, these symptoms were not severe and did not cause any serious safety problems.