The determination of the reference intervals for serum free thyroxine (FT4) and thyrotropin (TSH) is usually based on central 95 percentile intervals using subjects without detectable antibodies against thyroid peroxidase (TPO) or thyroglobulin (Tg). However, some subjects with extreme data over reference intervals are generally included. The study objective was to evaluate the reference intervals for FT4 and TSH using dif-ferent outlier tests. 1,007 Japanese subjects screened based on the National Academy of Clinical Biochemistry criteria in the United States participated in this study. Serum FT4, TSH, and TPOand Tg antibodies were mea-sured in all subjects. To make appropriate reference intervals, the Smirnov-Grubbus’ outlier test was taken for antibody-free subjects (Ab[-] S-G), and the conventional outlier rejection method rejecting the value out of ±3 standard deviation was taken for antibody-free subjects (Ab[-] STD) and all subjects (ALSTD), respectively. 12.8% of all subjects had either TPOor Tg antibodies in their serum. The 2.5^th and 97.5^th percentiles of refer-ence intervals of serum FT4 (ng/dL) and TSH (mU/L) were 1.03~1.66 and 0.51~5.14 in (Ab[-] S-G), 1.03~1.65 and 0.51~4.57 in (Ab[-] STD) and 1.03~1.66 and 0.51~4.67 in (ALSTD), respectively. FT4 in males were significantly and negatively correlated with age, and TSH was significantly and positively correlated with age (P<0.000001 and P<0.00001, respectively). There was a significant difference between the sexes in FT4 (P<0.00001) but not in TSH. The prevalence of hypothyroidism, subclinical hypothyroidism, suclinical hy-perthyroidism and hyperthyroidism were 0.2, 3.1, 2.3 and 0.3% (Ab[-] S-G), 0.3, 4.7, 2.3 and 0.3%(Ab[-] STD), and 0.3, 4.3, 2.3 and 0.3% (AL STD), respectively. This finding indicates that the conventional outlier rejection method is both convenient and appropriate to provide reference intervals for serum FT4 and TSH levels without regard to thyroid antibodies using large samples.