THE SEPARATION of patients with diabetes insipidus from individuals with normal ability to form concentrated urine has not always been satisfactory. Three diagnostic procedures have been utilized to make this differentiation. The Carter-Robbins test1 based on the observations of Hickey and Hare2 has been the procedure most frequently employed. The test is preceded by an overnight period of dehydration which in itself is dangerous in the small child with diabetes insipidus. The intravenous infusion is uncomfortable and catheterization of the bladder is frequently necessary to collect accurately timed urine samples. A second diagnostic procedure has been the parenteral administration of nicotine to stimulate antidiuretic hormone release.3 This has proven to be unreliable and uncomfortable for the patient. A water deprivation test employing the measurement of urinary specific gravity4 has been largely abandoned because the 24-hour period of dehydration required was hazardous in patients with diabetes insipidus.