An instrument was designed to determine relationships between constructs of the Expanded Health Belief Model and to identify characteristics of college students who successfully manage their diabetes. The Diabetes College Scale was developed to measure attitudes and behaviors pertinent to diabetes management and college life. It was tested for content validity, test-retest reliability, and internal consistency. Data were collected from college students using a cross-sectional design. Campus health care providers were invited via electronic mail to administer the survey to students with Type I diabetes. Ninety-eight questionnaires were mailed to interested providers, of which 86 (88%) were returned. Mean scores for attitude constructs, seven behaviors, and two outcomes were measured. Twenty-six experts established content validity. Instrument reliability was evaluated using paired t-tests, Cronbach's alpha, and correlation coefficients. Correlation coefficients and stepwise multiple regression analysis evaluated relationships among variables measured. Intention and emotional response were strong predictors of exercise, whereas health importance and intention were predictive of testing blood sugar. Situational factors and emotional response were substantial barriers to optimal diabetes self-care. College health care providers should address these areas in providing services to this population. Additional testing of the instrument is also recommended.