Isolates of Salmonella typhimurium, recovered over a 9-month period from a child with gastroenteritis, were characterized by biotyping, phage-typing and plasmid-profile analysis. Because the data from the different methods were discrepant, it was difficult to establish whether her infection was due to a single strain that had changed character in vivo or was due to recurrent infections with different, unrelated strains. Restriction-enzyme fingerprinting of the plasmids from the different isolates provided an explanation for the initial discrepancy and high-lighted a source of potential confusion in epidemiological studies.