The concentration of unextracted urinary arginine vasopressin (UAVP) was directly measured by high-sensitive radioimmunoassay (AVP-RIA Kit, Mitsubishi Petrochemical Co., Ltd.). Urine was diluted to eliminate interference of nonspecific substance without prior extraction. When urine aliquots were diluted in 4 to 32 fold in assay buffer, the relationship between UAVP concentration and dilution ratio corresponded exactly in a linear regression line. The elution pattern on Sephadex G-25 of UAVP immunoreactivity was identical with that of synthesized AVP. The AVP concentration in unextracted urine was not significantly different from that of extracted urine by Sep-Pak C18 column (Water Associates, Milford MA). The mean recovery of added AVP to urine specimens was 101.1 +/- 9.8% (mean +/- SD). The immunoreactivity of UAVP was not modified by either albuminuria (50 and 100 mg/dl) or glycosuria (1000 g/dl). Mean coefficients of variance between-assay and within-assay were 8.3% and 6.6% respectively. In normal subjects (n = 28), significant correlation was observed between UAVP concentration and simultaneously measured plasma AVP (r = 0.701, p less than 0.001). Moreover, AVP concentration in random urine was significantly correlated with AVP excretion in 24 hr-urine (r = 0.703, p less than 0.05, n = 9), and this suggested that random UAVP concentration may indicate daily UAVP secretion. In normal subjects, AVP concentration in random urine was widely scattered from 9.2 to 470.6 pg/mg Cr (89.5 +/- 76.4 pg/mg Cr, n = 211). In patients with diabetes insipidus (DI), UAVP concentration (1.6 to 13.0 pg/mg Cr, 6.94 +/- 2.77 pg/mg Cr, n = 25) was significantly lower (p less than 0.001) than that of normal subjects. UAVP concentration in a patient with primary polydipsia (43.2 pg/mg Cr) was not similar to that of ID but to that of normal subjects. UAVP concentration in 2 patients with SIADH was not more than that of normal subjects, indicating that random UAVP concentration is not suitable for detecting inappropriate AVP secretion. In this study, it is suggested that patients of random UAVP concentration below 13.0 pg/mg Cr should be recommended other intensive examination to diagnose DI, even though 2 normal subjects (0.9%) were incorrectly estimated as DI. In conclusion, radioimmunoassay of AVP in unextracted random urine is easy to sample and assay, and useful in screening polyuric patients.