Serum fructosamine activity was studied in 42 patients with type I (insulin dependent) diabetes mellitus and 30 non-diabetic volunteers as an index of blood glucose control. There was a significant correlation both between fructosamine and glycosylated haemoglobin values (r = 0.82) and between fructosamine and the fasting C peptide concentration (r = -0.81). Test results in 14 of the diabetics reflected the mean plasma glucose concentration calculated from 25 serial estimations in a single 24 hour period (r = 0.75; p less than 0.01) but not the mean amplitude of glycaemic excursion (r = 0.23; p greater than 0.05). Fructosamine concentrations measured in these multiple blood specimens did not change significantly throughout the day (mean coefficient of variation 4.1%) despite wide variability of the respective plasma glucose concentrations (mean coefficient of variation 36.2%). It is concluded that a single random serum sample analysed for fructosamine concentration provides a simple and reliable assessment of glucose homoeostasis in patients with type I diabetes mellitus.