Automated data processing and quality control of radioimmunoassays offer not only increased speed but also a more thorough and statistically rigorous analysis of results. An external quality assessment scheme for serum thyroxine, triiodothyronine and thyroid stimulating hormone (TSH) assays was performed in five nuclear medicine laboratories in Korea to compare with the assay performances of the World Health Organization Radioimmunoassay Program. The required radioimmunoassay kits were supplied through the International Atomic Energy Agency (IAEA). We have determined the weighted root mean squared error, and variance ratio as the indices of standard curve and also the average batch coefficient of variation (ABCV) as the parameters of response error relationship curve and precision profile. There was a good fit for the triiodothyronine assay, but 3 of 5 laboratories showed possible bad fit in the T4 and TSH assay systems. The ABCV was less than 5 percent for the T3 and T4 assay system, however for the TSH system, only 1 laboratory showed the ABCV value of less than 5 percent. We have also calculated the within batch variation (drift) and between laboratory variations.