The method developed by Vandamme (1994, Phys. Earth Planet. Inter., 85: 131-142) for the selection of the virtual geomagnetic pole (VGP) data set in terms of an optimum cutoff angle derived from analyses of synthetic VGP populations is applied to three Brunhes VGP data sets from central Mexico. The results are compared with those obtained by applying the criteria developed by McFadden (1980, Geophys. J. R. Astron. Soc., 60: 391-396), which assume that VGPs that are not part of a Fisherian distribution at a selected probability level (95%) are to be excluded from the paleosecular variation (PSV) analysis. The two methods give comparable results in the examples examined. Vandamme's method should be preferred for PSV studies, instead of those that use a given constant cutoff angle for data selection. Use of a constant cutoff angle results in higher PSV estimates.