The "normal range" is most useful as a standard of comparison for a given patient's laboratory result when it is derived from a homogeneous subpopulation closely matched to the patient for age, sex, and living habits. Some test values, notably urea nitrogen, cholesterol, and globulin, change with advancing age. The change seems to be due to the inadvertent inclusion of abnormal values in the preparation of the normal range. This is prabably because it is more difficult to select older individuals for optimum health. Many apparently healthy older persons are probably suffering from subclinical forms of degenerative disease. To increase sensitivity of case identification in mass-screening programs, it may be necessary to evaluate individual laboratory results against an optimum range of health rather than against the normal range as it is presently derived.