To evaluate the values of the thyroid autoantibody measured by radioimmunoassay (RIA) and compare it with hemagglutination method (HA) in the normal and the thyroid disease, data were obtained from total 61B persons; 236 healthy persons, 217 patients with Graves disease (including 113 patients with undertreated Graves disease'), 100 Hashimotos disease, 31 thyroid nodule, and 34 simple goiter. RSR kit made in England was used and could be detected t:o at least 3 U/ml. The positive rates of normal group were antirnicrosomal antibody (AMA) 31.8%, antithyroglobulin antibody (ATA) 44.5% by RIA and there was no considerable change in sex and age distribution. 1n Graves disease, the positive rates of AMA and ATA were 90.4, 76.9% by RIA, 85, 39% by HA. In Hashimotos disease, 94,91% by RIA, and 87,48% by HA, respectively. The autoantibody titer by RIA in thyroid autoimmune disease as welt as in normal group was more senisit.ive than that by HA, especially in ATA. There were linear relationships between the titer of RIA and that of HA in AMA of Graves disease and AMA and ATA of Hashimotos disease. There was no relationship among thyroid autoantibody, free T, index, TBII, and TSH. The titers of AMA and ATA were found to decrease in patients with Graves disease during the course of antithyroid drug therapv. Of the 236 normal subjects, thirty-seven (15.7%) had concentrations of above 7.5 U/ml in AMA, forty-four (18. 6%) above 9 U/ml in ATA. These values were considered as the upper limit for the normal range. In Graves disease, 82,7, 53.8% were above 7.5, 9 U/ml, respectively;1n Hashimoto's disease, HZ, 79% were positive. We conclude that RIA was more sensitve than HA in measuring the thyoird autoantibody, but we will study further more for determining the normal range and its interpretation.