Erwinia amylovora, the causal agent of fire blight, is genetically very homogeneous, and current methodologies provide insufficient or contradictory information about the probable dispersal routes of the pathogen. With the final aim to obtain specific and reliable molecular markers for different lineages of the pathogen, we studied the molecular basis of rep-polymerase chain reaction (PCR) polymorphism using seven different arbitrary primers to fingerprint 93 E. amylovora strains from different countries, including Spain. Polymorphism was very low, and was displayed by only 11 E. amylovora strains, which produced 22 polymorphic bands. Five of 11 polymorphic bands cloned contained DNA that was present in more than 85% of the strains, whereas six bands were due to DNA present exclusively in the strains producing the rep-PCR polymorphism. Also, five of the polymorphic bands were due to the possession of either the ubiquitous plasmid pEA29, of plasmid pEU30, which was exclusively found in strains from North America, or of a 35-kb cryptic plasmid, present only in 28 strains from Northern Spain. We designed primer pairs from several cloned polymorphic bands that allowed the specific identification of the strains producing the polymorphism. Our results indicate that rep-PCR is not adequate for constructing genealogies of E. amylovora, although the strategy illustrated here, as well as the designed primers, can be used effectively in epidemiological studies with this pathogen.