Radioimmunoassay techniques have been widely used for the measurement of serum thyrotropin concentrations. However, in spite of the great sensitivity of these techniques, some normal thyrotropin concentrations are still less than the minimum detectable amount. Data set in which contents below a fixed value are undetectable is referred to as Type 1 censored sample, and the maximum likelihood method can be applicable to define the normal range in this type of data. We have simultaneously measured serum thyrotropin concentrations obtained from 107 normal adult men by double antibody radioimmunoassay, and attempts were made to calculate the normal range by the maximum likelihood method. Distribution fitting of the censored data was determined by conventional chi-square tests, and it was found that the data fitted a normal log distribution. Although the serum thyrotropin concentration was undetectable (less than 0.625 microU/mL) in 14% of the subject, a normal range could be calculated. Using one-side tolerance limits for 95% coverage of the population with 90% confidence, we obtained the normal range of thyrotropin as 0.39 to 4.8 microU/mL with a mean value of 1.37 microU/mL, and predicted that 74.8% of undetectable thyrotropin values will fall within the normal range calculated above.