The aim of the present study was to determine whether venous blood samples can be used as an alternative to arterial samples in conscious and in anaesthetised dogs. Blood gas parameters (blood pH, PCO2, HCO3act-, BEecf, ctCO(2)) were measured in the femoral artery and in the cephalic vein of 12 dogs before (conscious dogs) and during profound anaesthesia induced by ketamine (15 mg/kg IM) after sedation with xylazine (2 mg/kg IM). Anaesthesia induced significant increases of all arterial values of gas parameters except for pH values which have significantly decreased. Similarly venous values were also significantly modified except for BEecf and HCO3act-. Arterial pH values were significantly higher than venous values whereas all the other parameters were significantly lower in arterial blood in conscious and in anaesthetised dogs. High positive correlations were evidenced between arterial and venous values of a given parameter in conscious dogs (the correlation coefficients were comprised between 0.861 and 0.947) as well as in anaesthetised dogs although they were less intense (the correlation coefficients ranged from 0.824 to 0.895). During anaesthesia, the highest correlation rate (r(2)> 80%) was obtained between arterial and venous pH values. In conclusion, venous blood gas values can predict arterial values in healthy dogs with sufficient accuracy whereas only venous pH values may be used during ketamine anaesthesia.