The within-clinic use of glycosylated haemoglobin (HbA1) and plasma fructosamine results in assessing blood glucose control and clinical management was compared in 1030 diabetic patients. The physician initially reviewed the patient with one randomly allocated measure (HbA1 or fructosamine) and completed a questionnaire concerning perception of blood glucose control, alteration to diet, alteration to medication, referral for diabetes education, and follow-up interval. The patient was then re-assessed using the second measure and the questionnaire repeated. Discordance rates for the study end-points, judged as binary outcomes, were: blood glucose control 15%; alteration to diet 7%; alteration to medication 9%; referral for education 3%; follow-up interval 4%. A significantly greater number of patients were rated as poorly controlled with HbA1 than with fructosamine (p less than 0.001) and were, in consequence, more frequently recommended alteration to diet and medication, referral for education and shorter follow-up interval; the rate of discordance for at least one of the management decisions was 16%. Multifactorial analysis showed that discordant management was dependent on the reviewing physician (p less than 0.001) and a history of cardiovascular disease (p less than 0.01); but neither type of diabetes, nor presence of nephropathy or variant haemoglobins, nor plasma glucose concentration, significantly influenced the likelihood of a discordance. Replacing HbA1 with fructosamine in the diabetic clinic may result in significant differences in the physician's perception of blood glucose control and in the management of patients.