The research used ventricular blood samples from 6 diabetic and 16 non-diabetic autopsied cadavers from 3 to 242 hours after death in Shimane prefecture in Japan in 1990-91. The fructosamine level is usually measured by the reduction of nitroblue tetrazolium (NBT) using a spectrophotometer. This assay was significantly disturbed by more than 1 gram per liter hemoglobin, which strongly reduced NBT by the catalytic action of the sulfhydryl and glycated groups. Glycated and total hemoglobin levels were then determined simultaneously to exclude the interferences. Total protein concentration was analyzed to eliminate dilutional effects by hemolysis. With these modifications, corrected fructosamine levels in samples containing not more than 10 grams per liter hemoglobin could be estimated. The assay of such hemolytic samples from the cadavers indicated higher fructosamine values in the diabetics than in non-diabetics and the postmortem degradations of the levels than previously reported. Tables, figures, and 26 references (Author abstract modified)