A constrained variational approach is used to derive self-consistent equations for determining the optimal, pH dependent charge state of all the protonatable groups in a protein. The approach uses a screened Coulomb potential (SCP) with a sigmoidal, distance dependent dielectric screening function for calculating the interaction energies between the charges in a protein. In addition, a formula is derived from the integral Born equation, using radially dependent permittivities, for calculating the electrostatic free energy of transferring a charged group from pure solvent to a protein in a dilute solution. The relationship between the approach and the Lorentz-Debye-Sack theory of polar solvation is outlined, and the method has been applied to calculate the pK’s of the titratable groups in bovine pancreatic trypsin inhibitor, hen egg white lysozyme, ribonuclease A, and ribonuclease T-1. The results are compared to recent calculations of pH dependent properties of these proteins that used the finite difference method for solving the Poisson-Boltzmann equation (FDPB). The comparison shows that the overall reliability of the present method is similar to that achieved by the FDPB approach but that the approach derived here is about 10(2)-10(3) times faster. The origin of this increase in computational speed is the calculational simplicity inherent in the use of a SCP and the fact that the method bypasses the need for numerically solving a second order differential equation for complex, many-body systems. The pH dependence of the pK is analyzed by comparing the pK(1/2) and the effective pK at pH 7. While the pK of most residues change only slightly, in some cases shifts up to 1.4 units were calculated, The SCP approximation is compared to the generalized Born (GB) treatment, and it is shown that the latter can be expressed in terms of an interaction that is damped by a sigmoidal screening function and a self-energy term. Comparison of interaction energies calculated from the two methods shows excellent agreement. In spite of this agreement and the reasonable overall root mean square deviation between calculated and measured pK, errors of more than one unit were found in several cases. The possible sources of these errors are analyzed and traced to the various approximations made in calculating the interaction and solvation energy contributions to the final pK. From this analysis several steps are outlined for improving these approximations that may help decrease the discrepancies between calculated and experimental results.