We studied the accuracy of three techniques for the rapid estimation of the blood glucose concentration with reagent strips in samples of blood from inpatients with diabetes mellitus. We observed that the visual inspection of Dextrostix gave results that were too inaccurate for this technique to be of practical clinical benefit. In contrast, when performed by registered nurses, both the Dextrostix-Eyetone method and the visual inspection of Chemstrips yielded results that were closely correlated with the measured serum glucose concentrations. With both of the latter techniques, less than 11% of the determinations varied from the actual serum glucose concentration by more than 20%. We conclude that both these techniques are sufficiently accurate to be of potential use in the home monitoring of blood glucose concentrations.