The value of a water deprivation test incorporating urinary arginine vasopressin (AVP) measurement was investigated in 13 patients with polydipsia and/or polyuria (complete central diabetes insipidus [CCDI] in four; incomplete central diabetes insipidus [ICDI] in five; secondary nephrogenic diabetes insipidus [NDI] in three; compulsive water drinking [CWD] in one) and a group of 25 control subjects (C). Urine samples were collected after water deprivation during sleep and the urinary osmolalities and AVP concentrations were measured. Analysis of the results of 104 urine samples from the 25 control subjects demonstrated a close correlation between urinary osmolality and AVP (r = 0.89, P < 0.001). After water deprivation during sleep, the respective mean maximal urinary osmolalities and AVP concentrations were: 127.4 +/- 34.4 mOsm/kg and 1.1 +/- 0.5 pg/mL in the patients with CCDI (14 samples, four children); 410.3 +/- 101.8 and 6.1 +/- 3.5 in those with ICDI (16 samples, five children); 348.7 +/- 71.2 and 100 +/- 45.1 in those with NDI (nine samples, three children); 541.5 +/- 143.5 and 43.6 +/- 33.2 in the patient with CWD (two samples, one child) and 898.8 +/- 186.3 and 97.4 +/- 50.4 in group C (54 samples, 18 children). Furthermore, the urinary AVP level relative to the osmolality in each patient varied depending on the AVP secretion status and renal concentrating ability. Each patient, except the one with CWD, could be discriminated from the normal subjects using this test. It seems that this test is easy to perform and useful for diagnosis and follow-up of patients with partial/complete posterior pituitary function defects and those with renal concentration impairment.