Vasopressin (VP) activity has been measured in urine for assessments of long-term changes in hormone release. Most laboratories employ radioimmunoassays (RIAs) and extraction of urine for research. It has been assumed that the specificity of the antiserums used and the extraction procedures would be adequate to safeguard against non-VP substances that would interfere with the RIA. New evidence is presented that this is probably not the case. High-performance liquid chromatography of unextracted urine or CG-50 extracts of urine clearly separates two major immunologically active peaks. One corresponds to native VP and the other migrates with the front. Different antiserums recognize the non-VP peak variably, but the VP peak the same. These activities were not separated by Sephadex G-25 chromatography. In addition to variability arising in the measurement of VP, the VP excreted may be influenced by variable renal clearance of the hormone. There is ample evidence to assume that not all plasma VP is filterable and that the proportions of bound and unbound VP vary. Furthermore, there is tubular metabolism and secretion of VP. The variability of these functions is unknown.