Normal ranges by usual definition are derived from subjects free of disease. Why do we not then have normal ranges for patients with common diseases such as diabetes mellitus? When a blood sample for thyroid hormone concentration determination is ordered on a diabetic patient, the responsible physician should be aware whether the diabetes itself alters the normal range. The problem, of course, is the many variables, such as use of medications, degree of diabetic control, and even possible familial differences in binding of thyroid hormone. To actually quantitate these variables is difficult. For example, women receiving birth control pills do reliably have an increase in thyroid-binding protein levels and a consequent increase in blood levels of thyroxine (T4) and triiodothyronine (T,) as well as a decrease in the T, resin uptake. Even though many physicians are aware of these qualitative changes, they do not have data available to quantitate them.