A sensitive and specific assay for glycated albumin (GALE) has been developed. Ten pL of serum is fractionated by affinity chromatography on a boranateagarose column and quantitated by radioimmunoassay (RIA). Assay linearity is 15.6-1,000 ng of albumin. Dose-response curves prepared from both glycated albumin and albumin standard are superimposable in the range of 31-500 ng, equivalent to 0.1-15% of GALB assuming total albumin concentration to be 50 glL. Recovery was 93 f 3%. Sensitivity is 16 ng. Precision (CV) is 14.3% with patient sample duplicates (n = 57) and 15.6% when assaying controls (n = 21). Our reference interval (RI) for GALE is 0.43-1.19% of total albumin. GALB levels are significantly elevated (p < 0.0001) in random samples from 37 diabetics. GALB correlates positively with both glycated hemoglobin (GHB) (r = 0.901) and serum glucose concentration (r = 0.706). In 10 diabetic patients followed serially with fasting and postprandial glucoses, GALB, and GHB, GALE was more sensitive than GHB in reflecting short-term changes in glycemic control. Our data demonstrate the value of GALB measurements when the assessment of glycemic control over a short period of time is important, such as in monitoring the results of changing therapy.