Analysis of urine proteins of some individuals with proteinuria by SDS-PAGE and silver staining revealed protein bands in urine which did not appear to be present in plasma. The bands migrated with apparent molecular weights of 260 000, 180 000, 110 000, 45 000, 40 000, 30 000, 24 000, 18 000 and 11 000. These bands were shown to be albumin polymer and fragments by using a polyclonal antibody to (a) immunoprecipitate radiolabelled urine proteins, and (b) identify bands blotted from SDS-PAGE gels onto nitrocellulose paper. The specificity of the polyclonal anti-albumin antibody was confirmed by using two mouse monoclonal antibodies raised against human albumin which, between them, recognized the same protein bands on nitrocellulose paper as did the polyclonal antibody. The results of these studies of albumin in human urine confirm that albumin exists as polymer and also show that albumin fragmentation occurs in urine. Fragmentation occurs by proteolysis of the albumin molecule both at sites within and outside disulfide loops. The predominant cleavage site appears to be approximately two-fifths of the distance from one end of the albumin molecule to produce disulfide-linked fragments of about 45 000 and 30 000 molecular weight.