In routine clinical practice and in research interviews like the Diagnostic Interview Schedule (DIS), symptoms attributed to physical illness are excluded from contributing positively toward psychiatric diagnoses. The maximal rate of underdiagnosis which results from this diagnostic rule was determined in 114 patients with diabetes mellitus. When symptoms attributed to diabetes (total = 296, mean = 2.6 per patient) were discounted, 256 diagnoses were made. Eighteen additional diagnoses (a 6.7% increase) were appreciated when these excluded symptoms were considered as psychiatrically significant; however, only two of these new diagnoses were given to patients previously without a psychiatric diagnosis. The authors concluded that the DIS is sensitive in its detection of psychiatric illness in diabetes and perhaps other physical illnesses as well.