In a comparative study, the PCR-based RAPD and ERIC fingerprint methods were evaluated for their resolving power to discriminate among 21 isolates of a natural Rhizobium meliloti population. PCR fingerprint patterns were analysed by using an automated laser fluorescent (ALF) DNA sequencer, thus allowing the automated on-line storage of data. Results obtained were compared to a classification system using insertion sequence (IS) fingerprinting. Both PCR fingerprint methods were comparable in their ability to resolve differences amongst Rh. meliloti isolates. Grouping of strains on the basis of their RAPD as well as their ERIC fingerprints correlated with grouping of strains according to their IS fingerprints. Moreover, strains displaying identical PCR patterns could be further differentiated according to their IS fingerprints, thus allowing a detailed insight into phylogenetic relationship among strains. The automated evaluation of strain-specific fingerprint patterns has the potential to become a valuable tool for studies of bacterial population genetics. Moreover, the rapid identification of single strains, e.g. pathogens in epidemiological studies seems feasible.