Researchers have sought to explain nonadherence to standard medical regimens by investigating a variety of sociodemographic variables, and, less often, by exploring variations between the health perspectives and life circumstances of the individual. While divergence between lay and professional perspectives on the etiology and treatment of chronic diseases, such as noninsulin dependent diabetes mellitus, may possibly account for the documented low rates of adherence to biomedical recommendations, health beliefs and activities are best understood as connected to an individual's personal history and circumstances. In order to evaluate the relationship between causal explanation of NIDDM and adherence, ethnographic interviews were conducted among 51 older (65+) women with diabetes and their physicians. We chose to investigate adherence to dietary recommendations because it represents one of the most challenging lifestyle modifications and is particularly important to maintaining glycemic control. The interviews involved ethnomedical and food frequency intake questionnaires and semi-structured interviews. Results reveal a stronger association between dietary adherence and etiological perspectives on diabetes than any sociodemographic factors, including ethnicity, education and income or other health belief factors. Informants suggested five categories that they believed were responsible for the onset of their NIDDM; poor past dietary practices (n = 22); familial tendency to have diabetes (n = 10); improper bodily functioning (n = 10); personal risk factors (n = 6); and currently being overweight (n = 3). Analyses indicated that those who implicate former dietary practices, currently being overweight, or having improper bodily functions were more likely to follow a standard recommended diet for individuals with diabetes. These findings also highlight the attempt by individuals with NIDDM to create 'stories' of meaning of their diabetes by linking their current management strategies for NIDDM with past practices and history. In addition, our results question the utility of the 'biomedical/alternative' labels.