Home health nurses provided individualized instruction in diabetes self-care within the home environment of 393 diabetic individuals. Each subject was randomly assigned to either the intervention (those receiving home teaching) or control (those not receiving home teaching) group. At 6 mo postenrollment, intervention subjects showed significantly greater self-care knowledge and skills than control subjects, although the actual differences between the two groups in terms of self-care skills were probably too small to have any practical meaning. The primary objective of the study, which was the reduction of the number of preventable diabetes-related hospitalizations (ketoacidosis, ketotic coma, nonketotic coma, insulin reaction, and diabetes out of control), was not achieved; no differences between the groups were noted after 12 mo of follow-up. Similarly, length of hospital stay, foot problems, emergency room and physician visits, and sick days were roughly equivalent in both groups during the follow-up year. These results suggest that, in the absence of concurrent changes in the health-care delivery system and strategies for influencing attitudes toward self-care, education alone is ineffective.