1. A method is described for the isolation and vascular perfusion in vitro of the mandibular gland of the rabbit. The perfusate is a physiological salt solution containing glucose as the only metabolic substrate.2. During perfusion with solutions containing acetylcholine, the gland secretes vigorously at a rate and in a manner similar to that seen in vivo. Although the gland becomes oedematous during perfusion, the extent of this oedema appears to have no influence on secretory ability: the perfused glands were capable of functioning for at least 4 h, and often for more than 6 h.3. Acetylcholine evoked a small secretory response at a concentration of 8 x 10(-9) mol l(-1) and a maximum response at 8 x 10(-7) mol l(-1). Eserine (2 x 10(-5) mol l(-1)) evoked secretory responses comparable to those evoked by acetylcholine in a concentration of 8 x 10(-9) mol l(-1). Secretion, whether unstimulated or evoked by acetylcholine or eserine, could be blocked completely by atropine.4. During prolonged stimulation with acetylcholine, the fluid secretory response declined rapidly over a period of about 15 min from an initial high value to a much lower plateau value. After 3 or more hours of stimulation, the secretory response began once more to decline, this time towards zero. If, before the second period of decline begins, stimulation is interrupted for about 30 min, the gland recovers its initial responsiveness to further stimulation with acetylcholine.5. The Na, K, Cl and HCO(3) concentrations and the osmolality of acetylcholine evoked saliva exhibited flow-dependency similar to that seen in vivo. The concentrations of Na and Cl, but not K and HCO(3), increased by about 25 mmol l(-1) during periods of prolonged stimulation with acetylcholine even though the salivary secretory rate was constant. The concentrations of K and HCO(3), but not Na and Cl, increased progressively as the concentration of infused acetylcholine was increased.6. Salivary protein secretion increased with increasing concentrations of acetylcholine to a greater extent than did fluid secretion. During continuous stimulation, the rate of protein secretion fell off much faster than the rate of fluid secretion.7. The beta-adrenergic agonist isoproterenol evoked a fluid secretory response only equal to about 5% of that evoked by acetylcholine, but still the response declined during continued stimulation. The electrolyte composition of isoproterenol-evoked saliva was vastly different from that evoked by acetylcholine, being particularly rich in K and HCO(3). The isoproterenol-evoked saliva was also extremely rich in protein so that the total protein secretion evoked by isoproterenol was much greater than that evoked by acetylcholine.8. The alpha-adrenergic agonist phenylephrine was without stimulatory effect on salivary fluid secretion and caused a reduction in the secretory response to acetylcholine. The drug had little or no effect on the electrolyte content of acetylcholine-evoked saliva and appeared to reduce its protein content.