Prevention of nosocomial pneumonia becomes an achievable goal only to the extent that the mechanisms of infection are known. It is likely that there is variation among microorganisms, patients, type of care and intensive care, and practices that determines the relative efficacy of preventive measures. Certain procedures appear to be universally required and include adequate reprocessing of ventilation equipment and infection control measures in patient care. In contrast, many factors may affect the role of the stomach as a cause of nosocomial pneumonia, including enteral feeding procedures and gastric acidity. Differences in results between studies of preventive measures may reflect uncontrolled practice factors, which make the measures either more or less important. Selective decontamination of the digestive tract is a measure with potentially serious consequences, primarily cost and microbial resistance, and should be undertaken with care probably in selected high-risk patients. Other recommendations such as the use of sucralfate, which is effective in prevention of bleeding without clear side effects, are probably worth instituting based on existing evidence. The primary need may be for better reprocessing of equipment or hand washing and sterile gloves, or it may be several changes at once. Certainly an array of preventive measures is necessary, and there is probably no single procedure likely to solve such a complex problem. Use of the collective findings of many investigators needs to be made and strategies applied to each patient and setting. There has been a marked increase in our knowledge of nosocomial pneumonia, and effective measures for prevention are available. Application of these measures widely should reduce the frequency of respiratory complications. The microbiologic tools to compare bacterial isolates have been developed, and the course of events preceding infection of the patient can be demonstrated. The role of equipment, environment, other patients, personnel, colonization sites, and other factors can now be examined. Future studies should control for the many known factors that may predispose to nosocomial pneumonia to make the results meaningful. This would include definition of infection, patient risk factors, identification of microorganisms, details of enteral nutrition, type of stress ulcer prophylaxis, exposure to antimicrobial agents, and institutional resistance patterns. In some studies, surveillance cultures and molecular epidemiology techniques would be required. Large controlled multicenter studies are necessary to determine the significance of the results of promising smaller studies.(ABSTRACT TRUNCATED AT 400 WORDS)