Corticosterone levels in blood may be used as a marker of stress in rodents, provided that the blood sampling procedure itself is non-stressful. Automated blood sampling equipment (Accusampler) allows blood sampling without any interference with the animal and might be useful as a tool for an on-line measurement of stress markers in blood. However, the impact of the blood sampling itself on the corticosterone levels in blood is unknown. The present study was designed to evaluate whether the frequency of blood sampling influences the plasma corticosterone levels in male and female rats. During anaesthesia, a catheter was placed in the jugular vein and attached to an Accusampler. Blood samples (200 microl) were withdrawn with a high (24 samples) or low frequency (3 samples) during a six-hour period immediately after the catheter insertion. The corticosterone levels in the plasma were quantified with ELISA. The corticosterone levels persisted at high post-operation concentrations when blood was collected frequently, while the levels steadily declined significantly during low-frequency sampling. The corticosterone levels were higher in female than in male rats, but the curves were similar. The present study elucidates the importance of considering the frequency of blood withdrawal during automated blood sampling. This parameter may have an impact on the experimental results when using blood corticosterone levels as a stress marker, but also during any in vivo study where blood is collected, since high corticosterone levels may affect the normal physiology of the animals.