The advances made in statistical methods to detect selection from DNA sequence variation has resulted in an enormous increase in the number of studies reporting positive selection. However, a disadvantage of such statistical tests is that often no insight into the actual source of selection is obtained. Finer understanding of evolution can be obtained when those statistical tests are combined with field observations on allele frequencies. We assessed whether the metallothionein (mt) gene of Orchesella cincta (Collembola), which codes for a metal-binding protein, is subject to selection, by investigating alleles and allele frequencies among European metal-stressed and reference populations. Eight highly divergent alleles were resolved in Northwest Europe. At the nucleotide level, a total of 51 polymorphic sites (five of them implying amino-acid changes) were observed. Although statistical tests applied to the sequences alone showed no indication of selection, a G-test rejected the null hypothesis that alleles are homogeneously distributed over metal-stressed and reference populations. Analysis of molecular variance assigned a small, but significant amount of the total variance to differences between metal-stressed and non-stressed populations. In addition, it was shown that metal-stressed populations tend to be more genetically diversified at this locus than non-stressed ones. These results suggest that the mt gene and its surrounding DNA region are affected by environmental metal contamination. This study illustrates that, in addition to statistical tests, field observations on allele frequencies are needed to gain understanding of selection and adaptive evolution. © 2007 Nature Publishing Group. All rights reserved.