Glomerular volume was estimated in 20 type 2 diabetic patients (age 64 ± 6 years, duration of diabetes 6 ± 5 years) compared with 14 sex- and age-matched controls, as well as in a group of 11 very long-term type 1 diabetic patients (age 61 ± 12 years, duration of diabetes 44 ± 11 years). One whole autopsy kidney was obtained prospectively, and a known fraction (~1/140) was sampled systematically and embedded in plastic (JB-4 glycolmetacrylate), thereby essentially eliminating shrinkage. Sections 15-μm thick were stained with periodic acid-Schiff. Mean glomerular volume was estimated on a random sample of glomeruli using the disector method. Frequency of glomerular occlusion and mean volume of open glomeruli was estimated. Mean glomerular volume was not different between type 2 diabetic patients and controls (5.3 ± 1.7 Mμ³/1.73 m² versus 5.3 ± 1.9 Mμ³/1.73 m²) nor was total glomerular volume or kidney weight. Frequency of glomerular occlusion was 4.8 ± 5.7% in controls, 8.9 ± 7.8% (p = 0.10) in type 2 patients, and 16.8% ± 20.7 (p < 0.05) in type 1 patients. In type 2 patients there was a correlation between frequency of glomerular occlusion and mean volume of open glomeruli (r = 0.44, p = 0.05), and the same tendency was seen in type 1 patients (r = 0.49, p = 0.12). By the present method the absolute level of glomerular volume was increased by at least a factor of two compared with previous studies. This illustrates the problems arising from shrinkage of tissue in paraffin and stresses the importance of using an unbiased stereological method. The lack of increase in total glomerular volume is in accordance with clinical findings of lack of glomerular hyperfiltration in type 2 patients, findings in contrast to those in type 1 diabetes. It is suggested that hyperfiltration per se is not the cause of glomerulopathy.