Age-matched reference values are generally presented with 5th and 95th percentiles as 'normal' reference range. However, they are mostly determined in relatively small groups, which renders this presentation inaccurate. We determined reference values for B-lymphocyte subpopulations in healthy children with the statistical method of tolerance intervals that deals far better with the relatively small numbers tested, and compared these to the cut-off values used in the currently used EUROclass classification for common variable immunodeficiency disorders (CVID) in children. CVID is a heterogeneous group of primary immunodeficiency diseases characterized by low serum immunoglobulin levels and inadequate response to vaccination. Disease-modifying heterozygous amino acid substitutions in TACI are found in around +/-10% of CVID patients. Interestingly, we found that age is the primary determinant of TACI-expression on B-lymphocytes, independent of switched memory B-lymphocyte numbers. Immunophenotyping of B-lymphocyte subpopulations is increasingly used to classify patients with CVID into subgroups with different clinical prognosis according to the composition of their B-lymphocyte compartment. These classifications were mainly developed with data obtained in adults. Because of the maturing paediatric immune system, they may not be equally applicable in children: our and other age-matched reference values show great changes in the composition of the B-lymphocyte compartment during development. Although the greatest changes in B-lymphocyte subpopulations occur below the age of 2 years, when the diagnosis of CVID cannot yet be made, it is likely that a classification developed in adults cannot be used to classify the prognosis of children.