Safranine and the cyanine dye, 3',3'-dipropylthiadicarbocyanine (diSC3-5), were examined as membrane potential probes in cytochrome c oxidase vesicles. The spectra of the vesicle-associated dyes resemble those of the same dyes in organic solvents, indicating that safranine and diSC3-5 probably dissolve in a hydrophobic region of the proteoliposomal membrane. This binding of safranine to proteoliposomes occurs with a dye-membrane dissociation constant in the micromolar range. The binding of safranine and of diSC3-5 to liposomes or proteoliposomes is accompanied by fluorescence enhancement. This enhanced fluorescence is quenched by respiration or by the establishment of a K+ diffusion potential by valinomycin (negative interior). An optimal dye/lipid ratio was required to secure maximum fluorescence quenching of the dyes, whether that quenching was active or passive. Calibrations of both the safranine and the diSC3-5 responses with K+ diffusion potentials were also affected by the dye/lipid ratio. At lower dye/lipid ratios, the calibration curve was linear at higher potentials but deviated from linearity at lower potentials. The converse was true at higher dye/lipid ratios. The non-linearity of the calibration curve at higher potential was attributed to a 'saturation' effect; it may also involve increased permeability of proteoliposomal membrane to protons. Destacking of dye at the lower dye/lipid ratio was probably responsible for the non-linearity of the calibration curves at lower potentials. When all these effects are taken into account, the steady-state value of delta psi generated during maximal proteoliposomal respiration was calculated to be between 140 and 160 mV (interior negative) when measured with either safranine or diSC3-5. We conclude that quantitative estimates of delta psi values can be made using these probes in cytochrome c oxidase reconstituted proteoliposomes provided that appropriate precautions are taken.