Most data concerning errors and accidents are from industrial accidents and airline injuries. General Electric, Alcoa, and Motorola, among others, all have reported complex programs that resulted in a marked reduction in frequency of worker injuries. In the field of medicine, however, with the outstanding exception of anesthesiology, there is a paucity of information, most reports referring to the 1984 Harvard-New York State Study, more than 16 years ago. This scarcity of information indicates the complexity of the problem. It seems very unlikely that simple exhortation or additional regulations will help because the problem lies principally in the multiple human-machine interfaces that constitute modern medical care. The absence of success stories also indicates that the best methods have to be learned by experience. A liaison with industry should be helpful, although the varieties of human illness are far different from a standardized manufacturing process. Concurrent with the studies of industrial and nuclear accidents, cognitive psychologists have intensively studied how the brain stores and retrieves information. Several concepts have emerged. First, errors are not character defects to be treated by the classic approach of discipline and education, but are byproducts of normal thinking that occur frequently. Second, major accidents are rarely causedby a single error; instead, they are often a combination of chronic system errors, termed latent errors. Identifying and correcting these latent errors should be the principal focus for corrective planning rather than searching for an individual culprit. This nonpunitive concept of errors is a key basis for an effective reporting system, brilliantly demonstrated in aviation with the ASRS system developed more than 25 years ago. The ASRS currently receives more than 30,000 reports annually and is credited with the remarkable increase in safety of airplane travel. Adverse drug events constitute about 25% of hospital errors. In the future, the combination of new drugs and a vast amount of new information will additionally increase the possibilities for error. Two major advances in recent years have been computerization and active participation of the pharmacist with dispensing medications. Further investigation of hospital errors should concentrate primarily on latent system errors. Significant system changes will require broad staff participation throughout the hospital. This, in turn, should foster development of an institutional safety culture, rather than the popular attitude that patient safety responsibility is concentrated in the Quality Assurance-Risk Management division. Quality of service and patient safety are closely intertwined.