BACKGROUND: Red blood cell (RBC) sorbitol has been implicated in the pathogenesis of organic complications of diabetes mellitus. W8 investigated RBC sorbitol level as an indicator of glucose control or diabetic complications, and also evaluated whether RBC sorbitol/plasma glucose ratio is an indicator of diabetic complications. METHODS: RBC sorbitol levels were measured in 43 healthy persons and 133 diabetes mellitus (DM) patients by enzymatic method. We also tested linearity, inter- and intra- assay precisions. Plasma glucose and Hb Alc were measured by hexokinase method and HPLC, respectively. Hospital records were reviewed. RESULTS: The intra- and inter-assay coefficients of variation of RBC sorbitol test are 8.7% and 28.5%, respectively. Linearity is good. The RBC sorbitol level(3.60+/-1.00 ug/mL) and RBC sorbitol/plasma glucose ratio (2.37+/-0.98%) in diabetic patients are significantly higher than those in normal control (1.69+/-0.43 ug/mL, 1.85+/-0.49 per mill), respectively(p<0.0001). We can't observe correlation between RBC sorbitol and Hb Alc in BM patients, but observe that in non-treatment DM patients. We also observed correlation between Hb Alc and glucose and reverse correlation between RBC sorbitol ratio and Hb Alc. We can't find significant relation between diabetic complications and RBC sorbitol or RBC sorbitol/plasma glucose. CONCLUSIONS: We suggest that the reference range of normal RBC sorbitol level and RBC sorbitol/plasma glucose ratio by enzymatic method are 1.69+/-0.86 ug/mL and 1.85+/- 0.98%,. These Ire significantly different from DM patients and may be useful in diagnosis of DM.