Unequal differentiation between two types of loci (allozyme and DNA markers) across a Mytilus hybrid zone has recently been claimed as evidence for direct selection on some llozyme loci. We provide here a counterexample: a noncoding DNA locus that exhibits as much differentiation as the incriminated allozymes do. The levels of genetic differentiation varied widely among both allozymes and noncoding DNA markers and no clear difference emerged between the two types of markers. This suggests that the strong interlocus variance in genetic differentiation has been confounded with a discrepancy between marker types as a result of an insufficient and unbalanced locus sampling. Heterogeneity in differentiation among neutral loci can be created by stochastic variance during the allopatric divergence preceding a secondary contact. In hybrid zones, a further source of variance is differential introgression among chromosomal regions after the secondary contact owing to the local influence of selected genes on more or less distant markers. However, the degree of differentiation alone gives no way to distinguish indirect pseudo-selection (a regular and ubiquitous feature of hybrid zones) from direct selection. More generally, we suggest that comparative neutrality tests based on discrepancies among marker types have to be applied with caution when the presence of semi–permeable genetic barriers to gene exchange is suspected.