Colonization of the south-western European mountain ranges is suggested to have predominantly progressed from the Iberian Peninsula eastwards, but this hypothesis has never been tested in a statistical framework. Here, we test this hypothesis using Androsace vitaliana, a high elevation species with eight mostly allopatric subspecies, which is widely but disjunctly distributed across all major south-western European mountain ranges. To this end, we use plastid and nuclear sequence data as well as fingerprint (amplified fragment length polymorphisms) data and employ Bayesian methods, which allow co-estimation of genealogy and divergence times using explicit demographic models, as well as hypothesis testing via Bayes factors. Irrespective of the ambiguity concerning where A. vitaliana started to diversify -- both the Alps and the mountain ranges of the Iberian Peninsula outside the Pyrenees were possible -- colonization routes were not simply unidirectional, but involved Pleistocene connections between the Alps and mountain ranges of the Iberian Peninsula bypassing the interjacent Pyrenees via long-distance dispersal. In contrast, the species' post-glacial history is shaped by regional gene pool homogenization resulting in the genetic pattern showing good congruence with geographical proximity in agreement with a vicariance model, but only partly supporting current taxonomy.