1. Fragments of rat submandibular gland were pre-loaded with 86Rb+, an isotopic marker of potassium transport, and rate constants for 86Rb+ efflux were determined during superfusion with a physiological salt solution. 2. In sodium-containing solutions acetylcholine evoked a rapid and immediate increase in efflux rate. After reaching a peak value, the efflux rate initially declined rapidly, but a second, slowly declining phase to the response was also evident. The response could be resolved into Ca2(+)-independent and Ca2(+)-dependent phases. 3. The basal efflux rate was elevated during superfusion with solutions in which sodium had been replaced with either lithium or N-methyl-D-glucammonium (NMDG+). Although lithium had a greater effect, which was absent under calcium-free conditions, addition of calcium to initially calcium-free, lithium-containing solutions did not affect the rate of efflux. 4. In the presence of calcium the response to acetylcholine was augmented during exposure to lithium-containing, sodium-free solutions but, in contrast, slightly inhibited when NMDG+ was used as a sodium substituent. 5. The transient, calcium-independent component of the response to acetylcholine was unaffected by exposure to lithium, whereas the calcium-dependent phase of the response was inhibited. 6. Responsiveness to acetylcholine was reduced during superfusion with a calcium-free, NMDG+-containing solution. The response normally observed when extracellular Ca2+ was subsequently elevated, in the continued presence of acetylcholine, was also inhibited. Sensitivity to acetylcholine was retained, however, when the tissue was initially exposed to a solution containing approximately 20 mumol l-1 Ca2+. The response was smaller than that evoked in sodium-containing solutions. 7. The use of lithium as a sodium substituent presents special problems, possibly related to the effects of this ion on the metabolic cycling of phosphatidylinositol-4,5-bisphosphate metabolites.