Computer simulations were used to investigate population conditions under which phylogeographic breaks in gene genealogies can be interpreted with confidence to infer the existence and location of historical barriers to gene flow in continuously distributed, low-dispersal species. We generated collections of haplotypic gene trees under a variety of demographic scenarios and analyzed them with regard to salient genealogical breaks in their spatial patterns. In the first part of the analysis, we estimated the frequency in which the spatial location of the deepest phylogeographic break between successive pairs of populations along a linear habitat coincided with a spatial physical barrier to dispersal. Results confirm previous reports that individual gene trees can show 'haphazard' phylogeographic discontinuities even in the absence of historical barriers to gene flow. In the second part of the analysis, we assessed the probability that pairs of gene genealogies from a set of population samples agree upon the location of a geographical barrier. Our findings extend earlier reports by demonstrating that spatially concordant phylogeographic breaks across ;ndependent neutral loci normally emerge only in the presence of longstanding historical barriers to gene flow. Genealogical concordance across multiple loci thus becomes a deciding criterion by which to distinguish between stochastic and deterministic causation in accounting for phylogeographic discontinuities in continuously distributed species.