We investigated the phylogeographical patterns of Lissotriton italcus, a newt endemic to the Italian peninsula, aiming to determine why hotspots of intraspecific diversity so ‘hot’. We found two main mitochindrial DNA lineages (net sequence divergence of 6.8% at two fragments of total length of 1897 bp): one restricted to part of the Calabrian peninsula (i.e the southernmost portion of the species range) and the other widespread throughout the rest of the species range. Both lineages, which had a parapatric distribution, showed evidence of further subdivisions, with an overall number of eight terminal haplogroups, most of whose times to the most recent common ancestors were estimated at the Late Pleistocene. Analysis of molecular variance suggested that partitioning populations according to the geographical distribution of these haplogroups can explain 97% of the observed genetic variation. These results suggest that L. italicus underwent repeated cycles of allopatric fragmentation throughout the Pleistocene, and that it likely survived the Late Pleistocene paleoenvironmental changes within eight separate refugia. Thus, the current hotspot of intraspecific diversity of L. italicus (within the Calabrian peninsula) has not been moulded by long-term stability of large populations but rather by multiple events of allopatric fragmentation and divergence. When compared with the patterns recently identified in other species, these results suggest that the occurrence of phases of allopatric divergence (eventually followed by secondary admixture) could be a common, albeit probably underrated feature in the history of formation of hotspots of intraspecific diversity. © 2011 The Linnean Society of London, Biological Journal of the Linnean Society, 2012, 105, 42–55.