Health-associated reference values are universally needed in clinical chemistry. The aim of this study was to establish the reference intervals of two populations from data obtained by the mass screening of newborn babies and to demonstrate how to determine 95% confidence intervals around the lower and upper limits of reference values from values that are not normally distributed. Biotinidase activities from Belgian (n=260) and Turkish (n=328) populations were measured by fluorometric assay and expressed as 1 IU (1 nmol/(dl min). Neonatal thyroid-stimulating hormone (nTSH) values from Belgian (n=4186) and Turkish (n=1663) populations were also measured by the solid phase two-site fluoroimmunometric assay, the results were given as muU/ml blood. Transformation of data was performed for each parameter. A parametric method was used for determination of reference values of biotinidase activity and the Belgian population was significantly higher than the Turkish population. nTSH reference values were evaluated by an exact non-parametric method, but approximate calculation based on the central limit theorem was also performed for confidence intervals around the reference limits. nTSH values of the Turkish population were found to be significantly higher than for the Belgian population. Rank numbers were found by an exact non-parametric method based upon the assumption of a binomial distribution. This study shows a procedure to define the rank numbers for n>1000 and to obtain reference values with 95% confidence intervals for lower and upper reference limits.