Estimations of HbA~ levels have been used to assess long-term glycaemic control in spontaneously diabetic BB/E rats. The degree of metabolic control achieved by once daily insulin injections and continuous insulin infusion by osmotic minipump was compared. Citrate gel electrophoresis of lysed erythrocytes, previously washed and incubated in 0.9% NaC1, gave accurate HbA~ values without interference from either abnormal Hb variants or labile glycosylation products. Over a 12 week period there was no significant difference in the mean random weekly plasma glucose concentrations between diabetic rats maintained on insulin injections or continuous infusion therapy. The HbA~ values in the injection-treated animals remained unchanged throughout the study period (mean _+ SEM = 5.1 + 0.1%). Diabetic rats treated by osmotic minipump showed a steady decline in values over the same period (4.1_+0.1%; p < 0.001 vs injected rats) but levels remained higher than those recorded in non-diabetic control rats (2.9 +0.01%; p < 0.001 vs pump-treated rats). These differences in HbAI were reflected in the plasma glucose values obtained during a 30 h glucose profile performed after six weeks of insulin therapy. Diabetic rats on injection therapy showed considerable diurnal variation in plasma glucose concentration (5.5-11.2 mmol/1; mean 8.9 _+ 0.5) but continuous insulin infusion eliminated the fluctuations giving a significantly lower mean glucose level over the 30h period (7.3+0.1mmol/1; p < 0.005). HbA~ levels show a poor correlation with random plasma glucose estimations (r = 0.43) but provide a simple and accurate assessment of long-term glycaemic control without the need for multiple 24 h glucose profiles.