In late 1989 a quality assessment scheme was introduced for glucose meters at 12 non-laboratory sites in a unit of management. The overall monthly imprecision of the meters varied from 3.4% to 17.1%, the highest coefficients of variation being recorded for glucose concentrations outside the range 3-20 mmol/l. In the same period 37% of results fell outside +/- 10% of laboratory set target limits; 13% fell outside +/- 20% of these limits. Participants have been advised of the unreliability of results outside the range 3-20 mmol/l. The main benefit of the scheme has been the improved confidence of users in the results obtained.