A total of 440 skulls of common shrews, Sorex araneus, from Germany and Europe (Croatia, Hungary, Austria, and Norway) were studied. The material represented six chromosomal races (Ulm, Laska, Drnholec, Mooswald, Jütland, and Abisko) assignable to the Western European and Northern European karyotypic groups. The race of a few samples was not determined. Twenty-one linear measurements were taken on skulls and mandibles and used in this study. Pearson correlations and multiple linear regressions were used to see the relationship of the cranial variables to altitude, latitude, the chromosomal race, and the geographic location. The results from the tested samples differed; the most negative correlations to latitude were found in the samples assigned to the Western European karyotypic group (WEK), the least negative ones in the samples of the Ulm race. These results indicate the converse of Bergmann's rule. But taking into consideration all the samples of the different karyotypic groups across Europe, the correlations to latitude included positive ones, which would indicate that Sorex follows Bergmann's rule in some of the variables. The studied material of different karyotypic groups could not be differentiated in discriminant analyses. The separation of the studied races within the WEK alone was slightly better, but about 30 % of ungrouped cases remained. Only the separation of the regional samples within one chromosomal race revealed better results but was still very different between the races. This indicates that within the races, regional differences might be strong enough for a separation of the samples and that within a karyotypic group, and even more so across karyotypic groups, regional differences conceal racial differences.