The present study describes the epidemiology of severe hypoglycemia and identifies patient characteristics or behaviors associated with severe hypoglycemia in patients with insulin-dependent diabetes mellitus (IDDM) participating in the Diabetes Control and Complications Trial (DCCT). Patients and Methods The DCCT is a multicenter randomized clinical trial designed to compare the benefits and risks of intensive therapy with those of conventional management of IDDM. The DCCT's feasibility phase demonstrated that intensive therapy, with the aim of achieving glucose levels as close to the non-diabetic range as possible, was accompanied by a threefold increase in severe hypoglycemia compared with conventional therapy. This report is based on the first 817 subjects who entered the DCCT, with a mean follow-up of 21 months. Results Two hundred sixteen subjects reported 714 episodes of severe hypoglycemia; 549 (77%) occurred in intensively treated subjects. The incidence of severe hypoglycemia in the intensive treatment group ranged from two to six times that observed with conventional treatment. Severe hypoglycemia occurred more often during sleep (55%); 43% of all episodes occurred between midnight and 8 AM. Of episodes that occurred while subjects were awake, 36% were not accompanied by warning symptoms. In intensively treated subjects, predictors of severe hypoglycemia included history of severe hypoglycemia, longer duration of IDDM, higher baseline glycosylated hemoglobin (HbA1c) levels, and a lower recent HbA1c. Multivariate analyses failed to yield predictive models with high sensitivity. Conclusions In the DCCT, intensive treatment of IDDM increased the frequency of severe hypoglycemia relative to conventional therapy. Intensive treatment may cause even more frequent severe hypoglycemia when applied to less selected and less motivated populations in the clinical practice setting. These findings underscore the importance of determining the benefit-risk ratio of intensive and standard therapy of IDDM. Cultural Specificity and Comparison in Psychiatric Epidemiology: Walking the Tightrope in American Indian Research Increasingly, the mental health needs of populations are measured using large-sample surveys with standardized measures and methods. Such efforts, however, rarely include sufficient number of smaller, culturally defined populations to draw defensible conclusions about their needs. Furthermore, without some adaptation, the standardized methods and measures may yield invalid results in such populations. Using a recently completed psychiatric epidemiology and services study with American Indian populations as a case example, this paper outlines issues facing epidemiologists working in such culturally diverse contexts. The issues discussed include the following: (1) persuading the scientific community and potential sponsors that work with distinct or culturally defined populations is important; (2) framing research questions and activities to meet the needs of communities; (3) defining a population of inference; (4) balancing the needs for comparability and cultural specificity; (5) maximizing scientific validity in light of the challenges in sample acquisition; and (6) developing and implementing data collection methods that uphold scientific standards but are also realistic given the context. The authors draw on their experiences—most recently in the American Indian Service Utilization, Psychiatric Epidemiology, Risk and Protective Factors Project (AI-SUPERPFP)—to illustrate these issues and suggest ways to address each. A goal of this paper is to challenge those invested in conducting culturally valid epidemiologic work in such populations to better articulate the nature of these efforts. Dietary Assessment and the Reliability of Nutritional Epidemiology Research Reports It is a pleasure to comment on the valuable article by Dahm