Although many researchers have reported clinical and laboratory parameters for prediction of remission in Graves' disease during or after anti-thyroid drug therapy, there is no reliable one to assure the complete remission. We prospectively examined a practical therapy with minimum maintenance dose of anti-thyroid drugs for prediction of remission in Graves' disease. Fifty-seven patients with Graves' disease were treated with anti-thyroid drugs at the initial dose of 30 mg/day of methimazole (MMI) or 300 mg/day of propylthiouracil (PTU). Then, doses were gradually decreased, and finally discontinued when the patients were able to maintain euthyroid (normal FT4 and TSH) for at least 6 months with the minimum maintenance dose (MMI 5 mg every other day or PTU 50 mg every other day). After discontinuation of drugs, FT4, FT3, TSH and TSH-binding inhibitory immunoglobulin (TBII) were measured every one to two months for the first 6 months and every 3-4 months for the next 18 months to confirm continuous remission. After 2 years of drug cessation, 46 (81%) of 57 patients were in remission and the other 11 patients had relapsed into thyrotoxicosis. At the time of drug discontinuation, the serum concentration of FT4, FT3 and TSH, titers of anti-thyroglobulin antibodies and anti-thyroid microsomal antibodies, goiter size were not different between the remission and relapse groups. At the time of drug cessation, the activities of TBII and thyroid-stimulating antibodies (TSAb) overlapped between the two groups, although they were significantly lower in the remission group than in the relapse group (p<0.01). Forty percent (4/10) of TBII positive patients and 71% (23/32) of TSAb positive patients continued to be in remission. On the other hand, thyrotoxicosis relapsed in 5 (11%) of 47 TBII negative and 2 (8%) of 25 TSAb negative patients. These data indicate that minimum maintenance therapy to keep euthyroid (normal FT4 and TSH) for 6 months is a practical measure for 81% prediction of remission in Graves' disease. The measurement of TBII or TSAb gave little additional information for predicting remission.