Fragile sites are frequent sites of chromosomal rearrangements. Human cells contain numerous fragile sites that differ remarkably in their propensity for instability (Glover 1984), yet we do not understand why a subset of fragile sites is more prone to instability and to rearrangement than the other fragile sites. We have used retrotransposon overdose (RO) yeast cells as a model system for fragile site stability; retrotransposon pairs are known to serve as sites of instability (Lemoine 2005), and RO strains contain many more single retrotransposon elements and retrotransposon pairs in a variety of genomic contexts. RO strains show an elevated level of instability under the same conditions as human fragile sites such as replication stress. When RO strains are subjected to treatment with methylmethanesulfonate (MMS) in combination with a temperature-sensitive polymerase-α allele (pol1-17) or deletion of the replication checkpoint protein MEC1, there is an increased incidence of chromosome rearrangement in RO strains relative to wild-type. While RO strains displayed a diverse complement of chromosome aberrations, a subset were recurrent among independent isolates, suggesting that particular chromosome loci were more susceptible to chromosome breakage than others. Genome sequencing of strains containing chromosome aberrations revealed that many aberrations were coincident with Ty1 and Ty2 retrotransposons, with particular retrotransposon elements more likely than others either to be present at rearrangement breakpoints or to be deleted follwing replication stress. When comparing the sites of rearrangement induced in pol1-17 cells and mec1 cells, there were few common chromosome aberrations, suggesting that different mechanisms of replication stress may induce rearrangement at distinct sites. The location of chromosome aberrations was also examined following experimental lab evolution. As expected, RO strains displayed both a greater number and a greater diversity of chromosome rearrangements following lab evolution. Due to the destabilizing effect of retrotransposons, we expected frequent loss of Ty elements following evolution, yet we discovered both new insertions of Ty elements as well as deletions; these deletions were just as likely to occur at single retrotransposon elements as at retrotransposon pairs. These data suggest that the arrangement of retrotransposons as pairs is itself insufficient to make a locus unstable, but rather that additional factors govern the varying stability of these repetitive regions.