Two of the commonly used probes for measuring membrane potential -lipophilic cations and the cyanine dye diS-C3(5 ) -indicated nominally opposite results when tetraphenylarsonium ion was added as a drug to suspensions of metabolizing Bacillus subtilis cells.[3H]-Triphenylmethylphosphonium uptake was enhanced by the addition, indicating hyperpolarization, yet fluorescence of diS-C3(5 ) was also enhanced, indicating depolarization. Evidence is presented that both effects are artifactual, and can occur without any change in membrane potential, as estimated by S6Rb+ uptake in lthe presence of valinomycin. The fluorescence studl es suggest that tetraphenylarsonium ion displaces he cyanine dye from the cell envelope, or other binding site, into the aqueous phase.The uptake characteristics of the radiolabeled lipophilic cations were quite unusual: At low concentrations (e.g., less than 10 gM for triphenylmethylphosphonium) there was potential-dependent uptake of the label to a stable level, but subsequent addition of nonradioactive lipophilic cation caused further uptake of label to a new stable level. Labeled triphenylmethylphosphonium ion taken up to the first stable level could be displaced by 10mM magnesium ion, whereas S6Rb+ uptake was unperturbed. Association of the lipophilic cations with the surface of de-energized cells was concentration-dependent, but there was no evidence for cooperative binding. This phenomenon of stimulated uptake in B. subtilis (which was not seen in Escherichia coli cells or vesicles) is consistent with a two-compartment model with access to the second compartment only being possible above a critical cation concentration. We * Work performed while on sabbatical leave from Department of Biology, Ben-Gurion University of the Negev, Beer-Sheva, Israel.tentatively propose such a model, in which these compartments are the cell surface and the cytoplasm, respectively.Triphenylmethylphosphonium up to 0.5 mM exhibited linear binding to de-energized cells; binding of tetraphenylphosphonium and tetraphenylarsonium was nonlinear but was not saturated at the highest concentration tested (lmM). The usual assumption, that association of the cation with cell surfaces is saturated and so can be estimated on de-energized cells, therefore leads to undercorrected estimates of cytoplasmic uptake in B. subtilis, and hence to overestimates of membrane potential. We describe a more realistic procedure, in which the estimate of extent of binding is based on a mean aqueous concentration related both to the external concentration and to the much higher internal concentration that exists in energized cells. Using this procedure we estimate the membrane potential in B. subtilis to be 120mV, inside-negative. The procedure is of general applicability, and should yield more accurate estimates of membrane potential in any system where there is significant potential-dependent binding.