A 10 percent systematic, random sample of all CO's yielded 115 respondents, a response rate of 82 percent. A questionnaire was designed for the ascertainment of correctional officers' knowledge, attitudes, and practices in selected health areas. Knowledge was defined operationally as the completion of courses in basic first aid, advanced first aid, or cardiopulmonary resuscitation (CPR), the administration of medications, or as emergency medical technicians. Attitudes were measured with a Likert Scale in two major areas: inmates' access to care and relations with health workers and inmates. Practices were measured by asking CO's their frequency of refusal to let inmates attend sick call on unscheduled days during the week preceding the survey, their frequency of use of first aid and CPR during their careers, and the frequency of referrals of inmates with mental illness during 1 month prior to the survey. A majority of CO's had basic first aid training, but 31 percent or less had training in advanced first aid, CPR, or other skills. A minority of CO's were uncertain that other CO's would use CPR for inmates who needed it. Low ranking CO's were less likely than higher ranks to be certain about the inadequacy of frankly inadequate mental health care and were more likely to feel that CO's should decide if inmates are to attend sick call. Rural CO's were more likely to perceive inmates who complained of illness as faking. CO's used first aid once every 2.6 years, on the average, but made referrals of mentally ill inmates about 2.5 times a year. A large amount of consultation occurred before permitting inmates access to sick call on unscheduled days. Findings suggest that the basic training of CO's should identify the strengths and weaknesses of the medical and mental health systems, so CO's are not left to personal experience in determining the types of problems for which to refer inmates. Tabular data and six references are provided.