Historically, blood samples have been taken by puncture of the orbital sinus under light ether anaesthesia. In the last decade, there was concern over the use of ether as an anaesthetic. Furthermore, in recent years, more attention has been focused on the need to improve current laboratory practices for the bene¢t of animal welfare. Exposure for 2 min to a CO 2 /O 2 mixture produced acute hyperkalaemia and changes in a range of haematological parameters. Hyperkalaemia in some individuals reached pathological levels when compared to values normally obtained from ether-anaesthetised animals. Together with published data, our ¢ndings indicate that the use of a 70:30 mixture of CO 2 /O 2 as an anaesthetic in rats is questionable. With iso£urane, the only consistent ¢ndings following exposure for 5 minutes were a slight downward trend in the red blood cell parameters and in potassium, and an increase in glucose. These changes probably represent an e¡ect of prolonged exposure to iso£urane. Therefore, to avoid variation in these parameters, the duration of exposure should be limited to 3 min. When comparing sublingual to retro-orbital blood sampling after 2 min exposure to 4% iso£urane, no major di¡erences were found in the plasma chemistry parameters examined. However, interference with plasma amylase levels following sampling from the sublingual vein has been suspected in one toxicity study. The acute e¡ect of blood removal over 24 h on the main haematological parameters consisted of decreases in red blood cells (RBC), haemoglobin (HGB) and haematocrit (HCT), which were similar for blood removal exceeding 7.5% but below 15% of circulating blood volume, and were positively related to the amount of blood loss beyond 15%. There were no biologically signi¢cant e¡ects on these parameters below 7.5%. Time to recover from the e¡ects of the bleed was also proportional to the volume taken, and was estimated to range from 48 h for amounts between 5% and 7.5%, 12 days for 7.5% to 20%, and 19 days for amounts above 20 %. The collection of up to 20% of the circulating blood volume over 24 h did not a¡ect the welfare of rats, as indicated by the absence of mortality and clinical signs and the lack of e¡ects on body weight and food consumption.