A 6-month study of the effects on blood glucose control of computer-aided analysis of glucose self-monitoring results was performed. Eighteen clinically insulin-dependent patients were ranked in order of haemoglobin A1 and randomly allocated in consecutive pairs to either a conventional group, using diaries to record self-monitoring meter-read results, or a memory group using meters with inbuilt memories which were linked to a computer at each visit for the analysis of self-monitoring results. There was a significant decline in haemoglobin A1 (p<0.001) and fructosamine (p=0.002) over the 6-month period in the whole population but there was no difference between the study groups. The absolute or percentage changes in haemoglobin A1 and fructosamine did not correlate significantly with any of the patient characteristics examined. The memory system was popular with the patients. At present no benefit in terms of blood glucose control from such systems has been demonstrated.