The ATT39 scale was developed as a norm-referenced measure of emotional adjustment in diabetic patients. Scores on three parallel forms of the parent scale changed in response to educational intervention, and the change in scores was predictive of subsequent improvement in metabolic control. We describe further reliability and validity studies with six factorially derived subscales of the ATT39, which measured perceived levels of stress, adaptation, guilt, alienation, illness conviction, and tolerance for ambiguity. Internal consistency (Cronbach α) of the unweighted total score was 0.78, and the Guttman lower bound estimate of reliability was 0.86. The test-retest reliability of the total score varied from 0.70 to 0.87, over intervals of 2 wk, 3 mo, and 6 mo, and reliability coefficients for the six factor scores averaged 0.56. ATT39 factor scores, in 134 insulin-dependent diabetes mellitus (IDDM) and 166 noninsulin- dependent diabetes mellitus (NIDDM) patients, were correlated with scores on the Cattell 16 personality factor questionnaire and the locus of control of behavior scale (LCB). In IDDM, age was related to better adaptation, increased feelings of guilt, and a more cooperative attitude to staff and treatment. In NIDDM, age was associated with increasing resignation to a conviction of chronic illness and less tolerance for the ambiguities involved in diabetes. Intelligence was correlated with less guilt and more tolerance. Anxiety was associated with significant diabetes-related stress, regardless of treatment, and with poorer adaptation and guilt in NIDDM. An external LCB was related to increased stress and guilt. The results confirm that emotional adjustment in diabetes involves dynamic interactions among feelings that are relatively stable over periods up to 6 mo and that relate meaningfully to other aspects of personality functioning.