In a laboratory X-ray powder diffraction study, the evaluation of the patterns of three Bragg±Brentano powder diffractometers with different monochromator geometries has been undertaken. For the measurements on each diffractometer, the standard reference material SRM 640 (silicon) and the corundum samples SRM 674a and SRM 1976 have been used. In each case, the peak pro®les were ®tted with a split Pearson VII function and the FWHM (full width at half-maximum) parameters and exponent m were determined for the left (lower 2) and the right (higher 2) sides of the Bragg peaks. It was found that there is a strong dependence of both the FWHM and the exponent m on the diffraction angle for the two con®gurations that included monochromators, whereas nearly constant values of m were found for the geometrically simplest diffractometer working without a monochromator. Finally, the two components of the Cu K doublet show systematically different peak pro®les. There is a clear difference not only concerning the FWHM, which becomes more obvious at higher 2 values, but also in the course of m with respect to the diffraction angle for the left and the right tails of the powder re¯ections. This is the main reason for the dif®culties in K 2 stripping and also in single-line-pro®le analysis when using the K doublet. Therefore, it is not surprising that this phenomenon, which can be explained by Heisenberg's uncertainty principle, does affect the reliability (represented by standard R values) of structure re®nement by the Rietveld method.