We made fluorescence, electron paramagnetic resonance (EPR), electrophoretic mobility, and ionizing electrode measurements to study the effect of the monovalent lipid phosphatidylinositol (PI) and the trivalent lipid phosphatidylinositol 4,5-bisphosphate (PIP2) on the electrostatic potential adjacent to bilayer membranes. When the membranes were formed from mixtures of PI and the zwitterionic lipid phosphatidylcholine (PC), the Gouy-Chapman-Stern (GCS) theory described adequately the dependence of potential on distance (0, 1, 2 nm) from the membrane, mole % negative lipid, and [KCI]. Furthermore, all EPR and fluorescence probes reported identical surface potentials with a PC/PI membrane. With PC/PIP2 membranes, however, the anionic (coion) probes reported less negative potentials than the cationic (counterion) probes; the deviations from the GCS theory were greater for the coions than the counterions. Discreteness-of-charge theories based on the Poisson-Boltzmann equation incorrectly predict that deviations from the GCS theory should be greater for counterions than for coions. We discuss a consistent statistical mechanical theory that takes into account three effects ignored in the GCS theory: the finite size of the ions in the double layer, the electrical interaction between pairs of ions (correlation effects), and the mobile discrete nature of the surface charges. This theory correctly predicts that deviations from GCS theory should be negligible for monovalent lipids, significant for trivalent lipids, and greater for coions than for counterions.