Hospitalized patients are at unusually high risk of infections, and furthermore, the hospital environment favors the acquisition of resistance to antimicrobial agents, complicating the treatment of nosocomial infections due to drug-resistant pathogens. The prevention of nosocomial infections, based on a surveillance system as an essential element of an infection control program, is the only way to reduce morbidity and mortality. A typing method for strain clonality permits the infection control program to confirm the association between infected patients and the reservoir for the microorganisms of interest and to determine modes of transmission, because the mode of transmission or reservoir may not be the same for multiple strains of a bacterial species. An ideal method of subtyping bacterial isolates from a given species should be simple, rapid, sensitive and discriminatory. Traditionally, once bacterial isolates from an outbreak have been determined to be of the same species, further evaluation for similarity or relatedness has been based on phenotypic methods. Biotyping, serotyping, phagetyping, and antibiotype determination are not always adequately sensitive to distinguish unrelated strains with similar phenotypes. Within phenotypic methods, multilocus enzyme electrophoresis is a powerful tool but because of its complexity it is not likely to become widely available for study of local outbreaks of bacterial infections. In recent years, molecular genetic methods, including plasmid profile, genomic restriction fragment length polymorphism analyses by conventional electrophoresis or by pulsed-field gel electrophoresis, single chromosomal gene polymorphism by DNA hybridization or by PCR amplification, ribotyping, and genomic fingerprinting generated by repetitive element sequence-based polymerase chain reaction, have been useful in evaluating endemic infections and outbreaks of a variety of nosocomial pathogens. For epidemiologic studies, genotyping systems based on defined chromosomal genes or whole DNA polymorphism provide significant advantages over plasmid analysis. Among plasmid-non-based genotypic methods, the choice depends on i) the examination of how much discrimination the method can add for the epidemiologic investigation, ii) the resources available to the laboratory, and iii) the level of expertise of the personnel involved in the testing because, until standardized rules of interpretation are published, the same data may be interpreted in different ways by different investigators. Finally, biotyping, serotyping, and antibiotype determination remain an appropriate first step for the evaluation of apparent outbreaks with the caveat that different strains with the same phenotypic properties may exist concurrently within the same environment. Genotyping may be a second level of analysis to evaluate relatedness of bacterial strains, because the use of molecular biology techniques should support an epidemiologic investigation rather than initiate it.