The accuracy and outcome of capillary blood glucose (CBG) monitoring as routinely performed by nursing staff were assessed. The sample consisted of 160 readings conducted by 93 nursing staff members in four hospitals; 19% of the readings deviated from simultaneous laboratory results by greater than 20%, and deviations resulted in altered responses to standing orders in 26 patients (17%). There was no statistically significant difference between the total variation attributed within and between nurses, possibly indicating that all nurses should be given the same intensity of follow-up training rather than targeting those who perform incorrectly on sample tests. Nurses in the one hospital that required certification before CBG monitoring had significantly less deviation from the laboratory standard than the other three hospitals. Although these data do not provide direct evidence that the certification program increased accuracy, this seems a logical conclusion. More study is needed to determine the most cost-effective type of training and follow-up.