Subtotal thyroidectomy for Graves' disease sometimes leads to hypothyroidism or relapse during long-term follow-up in a significant proportion of patients. Factors predictive of postoperative hypothyroidism after subtotal thyroidectomy are not known. The objective of this study was to determine the relation between clinical features and expression of transcripts associated with thyroid hormone synthesis in resected thyroid tissues of patients with Graves' disease. Thyroid tissues were obtained from 65 patients with Graves' disease who underwent subtotal thyroidectomy. Expression of mRNAs from thyroglobulin (Tg), TSH receptor (TSHR), thyroid peroxidase (TPO), sodium/iodide symporter (NIS), and the Pendred's syndrome (PDS) genes were analyzed by quantitative reverse transcription-polymerase chain reaction. Uni- and multivariate analyses were performed to identify for postoperative hypothyroidism. We detected significant correlations between the NIS mRNA level and levels of free T(3) (fT(3)) and free T(4) (fT(4)) and between the Tg mRNA level and goiter weight before initial drug treatment. Mean levels of expression of all five mRNAs were significantly higher in patients who did not require L-thyroxine replacement therapy than in those who required replacement therapy at 6 months after surgery. In patients who did not require replacement therapy, a significant correlation was found between NIS mRNA expression and fT(4) levels. Univariate analysis revealed that decreased NIS mRNA expression (NIS/PGK<1.69) and low TBII levels before initial treatment were significant of postoperative hypothyroidism. Multivariate analysis showed decreased expression of NIS mRNA (NIS/PGK<1.69) to be an independent risk factor for L-thyroxine replacement after surgery (risk ratio, 3.26, confidence interval, 1.36-9.08, p<0.01). NIS expression reflects the level of thyroid hormone synthesis in Graves' disease patients. Evaluation of NIS mRNA expression in thyroid tissues may help determine prognoses of Graves' disease patients, and therefore an appropriate treatment can be determined for each patient.