Abstract
Parametric statistical methods may yield a distorted picture of the relationship between a laboratory quantity and age. A non-parametric method iteratively combining data of consecutive age groups was developed. At each step the two age groups with the minimum difference between the locations of the quantity distributions are combined, provided the locations and the dispersions do not differ significantly on specified levels. Finally, reference intervals are computed for the resultant age groups. The advantage of the method is illustrated by its application on existing data.