A formal evaluation of the Sarstedt capillary tube system for blood glucose sampling was undertaken. Results obtained with the method were found to be highly reproducible (coefficient of variation = 4.7%) and accurate (mean±SD difference between this system and a reference system =-0.14±0.45 mmol/l). A decline in glucose values by 10% occurred when they were stored for 48 h before analysis, and this decline persisted in samples kept at room temperature, but was abolished by storage at 4 °C.