A double-antibody radioimmunoassay for the insulin-degrading enzyme, glutathione-insulin transhydrogenase (GIT), has been developed with the use of rabbit antiserum against human liver GIT and [125I]-GIT. The method can determine as little as 32 fmol of GIT, thus allowing measurements in needle tissue biopsy samples and in plasma, which have not been possible with previous enzymatic procedures. Relative competition in the radioimmunoassay by unlabelled GITs purified from other sources are in agreement with homologies in GITs previously found using the enzymatic assay. No competition was observed with pork insulin, bovine ribonuclease, human albumin or human gamma-globulin, indicating that the radioimmunoassay is highly specific for GIT. Similar competition curves were observed for native GIT; active, reduced GIT; or for the inactive, S-(ethylsuccinimido) derivative of GIT. The radioimmunoassay thus measures total (active + inactive) GIT and permits determinations in the presence of materials which react with the active site and render the enzymatic methods unusable. Radioimmunoassay of plasma and extracts of liver, muscle and adipose tissues from diabetic and non-diabetic subjects showed parallel competition curves with standard purified human GIT indicating that GITs of non-diabetic and diabetic persons are immunologically very similar or identical. Concentrations of GIT in plasma determined by radioimmunoassay were significantly higher in diabetic than those in non-diabetic subjects (1620 +/- 80 versus 1070 +/- 30 fmol/l, p less than 0.001). Tissue GIT levels found by the radioimmunoassay as well as by the enzyme assay, both in non-diabetic and diabetic subjects, were highest in the liver, intermediate in the adipose tissue and lowest in the muscle.