Serum fructosamine, albumin and plasma glucose were studied in eight patients during recovery from diabetic ketoacidosis, and in eight patients with "decompensated" diabetes without acidosis. In the ketoacidotic group, serum fructosamine had fallen significantly by a mean of 12% by 8h, and fell further during the remainder of the study. A smaller but significant fall in fructosamine (mean 6.1%) was seen in the decompensated group after only 18 h, with again a further fall thereafter. These changes in serum fructosamine were accompanied by decreases in serum albumin. However, when fructosamine was corrected by calculating a fructosamine/albumin index (FAI) (fructosamine X 100/albumin), the FAI did not change significantly in either group until a small reduction was noted in the ketoacidotic group at 5 days, which might reasonably be expected as an index of intermediate-term glycaemia. Therefore minor fluctuations in albumin levels seen in diabetic patients can affect fructosamine, and correction may be be advisable for the test to be a valid measure of glycosylated serum proteins.