The IgG subclass protein concentrations in sera from 200 normal subjects were determined independently in two laboratories, using the same technique (radial immunodiffusion), the same subclass-specific antibodies and the same calibrator. The coefficients of correlation (r_s) between IgG subclass concentrations determined in the two laboratories were 0.487, 0.883, 0.928 and 0.926 for IgG1, IgG2, IgG3 and IgG4, respectively (p<0.0005 in all four cases). By the chi-square test for goodness of fit, the frequency distributions observed in the two laboratories were found to differ significantly for all four subclasses (p<0.01). In one laboratory, the distributions of IgG1 and IgG2 were not significantly different from normal distributions, whereas the distributions of IgG3 and IgG4 deviated significantly. In the other laboratory, all four subclass distributions were significantly different from normal distributions. In the first laboratory, the IgG2 concentrations were log-normal distributed, whereas in the second, IgG1, IgG2 and IgG4 concentrations were log-normal distributed. We conclude that use of identical reagents does not ensure identical frequency distributions. This finding emphasizes the need for standardization of the measurement technique too. Furthermore, we argue that at present, intralaboratory reference intervals for IgG subclass protein concentrations are necessary. The reference intervals should be based on non-parametric statistics. The subclass of 106 monoclonal IgG proteins, which were demonstrated by agarose gel electrophoresis, was identified by RID for 89 samples. The subclass of the remaining 16 M-components was readily determined by qualitative immunoelectrophoretic analysis using the same subclass-specific antibodies.