Plasmid profiling was used to determine the variability of normal flora isolates ofStaphylococcus epidermidisin order to evaluate the usefulness of plasmid profiling for identifying pathogens. Fifteen hospital staff members and patients repeatedly had cultures taken from the hands and nares, and multiple isolates were examined for plasmid profiles.S epidermidisisolated from the nares of 15 neonates were also examined. The total number of isolates examined for plasmid profiles was 726. Repetition of profiles was common among the different isolates from a single sampling (one swab). The frequency of re-isolating similar profiles on different days varied from 7% to 13%. Simultaneous isolation of similar profiles from nares and hands on the same individual varied from 0% to 11%, the percentage being lower for personnel. Isolation of the same plasmid profile from different individuals occurred only twice and resulted in an assignment probability ofPa= 0.002 for isolates obtained from different individuals. Significantly more isolates from nares contained plasmids (97%) compared with isolates from hands (89%).Patients who had two or more isolates of coagulase-negative staphylococci with similar profiles were judged, clinically, to have infections in 12 of 13 cases. However, the likelihood of re-isolating anS epidermidisstrain with a similar plasmid profile twice from the same person at different times was sufficiently high to prevent plasmid profiling from being used as an absolute criterion for infection.