Fifty-one clinical isolates of Listeria monocytogenes (15 isolates from two outbreaks and 36 epidemiologically unrelated isolates) were typed by conventional serotyping, ribotyping (RT), pulsed-field gel electrophoresis (PFGE), and arbitrarily primed PCR (AP-PCR). Serotyping was unable to distinguish between related and unrelated strains of L. monocytogenes. Each of the three molecular methods showed excellent typeability and reproducibility. Restriction with EcoRI and PvuII gave 16 and 23 RT patterns, respectively. Restriction with ApaI or SmaI generated 22 and 26 PFGE profiles, respectively. ApaI profiles were easier to interpret, with 10 to 15 bands each, while SmaI profiles had 15 to 20 bands each. AP-PCR with two different primers yielded 29 and 31 randomly amplified polymorphic DNA patterns, respectively. Strains from the same outbreak shared concordant patterns by each of the three methods. Of the three techniques evaluated, RT was the least discriminating and could not distinguish between strains from the two outbreaks. The abilities of AP-PCR and PFGE to differentiate between strains were comparable. However, AP-PCR was more rapid and easier to perform. We conclude that the DNA profiles generated by either AP-PCR or PFGE can be used to differentiate outbreak strains from epidemiologically unrelated strains and to clearly identify unrelated strains as being distinct from one another. We recommend that at least two independent primers be used for AP-PCR typing in order to improve its discriminatory power.