We have compared five different modes of anesthesia (iv and ip pentobarbital sodium, ether, CO2, and cervical dislocation) with respect to their effects on liver glycogen, liver adenosine 3',5'-cyclic monophosphate (cAMP), blood glucose and lactate, plasma corticosterone, norepinephrine, and epinephrine in resting rats and in rats run on a treadmill at 26 m/min for 30 min. Ether, CO2, and cervical dislocation were found to be unsuitable due to the marked elevation in plasma catecholamines seen in both resting and exercising rats. Injection of pentobarbital sodium ip required an average of 8 min before onset of surgical anesthesia as opposed to less than 5 s for iv pentobarbital. Exercising rats anesthetized with ip pentobarbital showed markedly lower plasma catecholamines compared with rats given iv pentobarbital. Hepatic cAMP increased in response to exercise in all groups except the ip pentobarbital group. This is most likely due to the long delay between the end of the exercise and freezing of the liver in the ip pentobarbital-anesthetized animals. We conclude that iv injection of pentobarbital is the most suitable method of anesthesia for obtaining accurate measurements of plasma stress hormones, substrates, and metabolites and of hepatic cAMP and glycogen in resting and exercising rats.