The issue of sampling sufficiency is too infrequently explored in phylogeographical analysis, despite both theoretical work and analytical methods that stress the importance of sampling effort. Regarding the evolutionary pattern of reciprocal monophyly, both the probability of recovering this pattern and the possible inferences derived from this pattern, are highly contingent upon the density and geographical scale of sampling. Here, we present an empirical example that relates directly to this issue. We analyse genetic structure in the southern Appalachian spider Hypochilus thorelli, using an average sample of 5 mitochondrial DNA (mtDNA) sequences per location for 19 locations. All sampled sites are reciprocally monophyletic for mtDNA variation, even when separated by geographical distances as small as 5 km. For populations separated by greater geographical distances of 20–50 km, mtDNA sequences are not only exclusive, but are also highly divergent (uncorrected p-distances exceeding 5%). Although these extreme genealogical patterns are most seemingly consistent with a complete isolation model, both a coalescent method and nested cladistic analysis suggest that other restricted, but nonzero, gene flow models may also apply. Hypochilus thorelli appears to have maintained morphological cohesion despite this limited female-based gene flow, suggesting a pattern of stasis similar to that observed at higher taxonomic levels in Hypochilus.