The initial step of the dephosphorylation of phosphotyrosine by the protein tyrosine phosphatases (PTPases) requires an active-site cysteine residue to be in its deprotonated (thiolate) form so as to make a nucleophilic attack on the phosphate to form a phosphoenzyme intermediate. At the same time, an active-site aspartic acid residue must be in its protonated form to serve as a general acid donating a proton to the leaving tyrosyl moiety. The protein must therefore provide an unusual electrostatic environment in the active site to maintain these protonation states in the free enzyme and in the Michaelis complex with the substrate, whose phosphate group is in the -2 charged form. We present electrostatic calculations for the pK(a) of the ionizable side chains of four different PTPases. The study includes Michaelis complexes with the substrate, as well as free enzymes. The calculations consistently predict that in the neutral to mildly acidic pH range, the active-site cysteine residue is deprotonated and the aspartic acid residue is protonated, as required for optimal catalysis. This prediction is made even for the Michaelis complexes. Good quantitative agreement with measured pK(a) values of the aspartic acid residues are obtained, while the calculated pK(a) values of the cysteine residue are generally much lower than experimental estimates, suggesting that the protonation of the (normally unprotonated) cysteine residue may induce a conformational change in the active site. An analysis is made of the electrostatic factors responsible for maintaining the functionally essential protonation states. The phosphate-binding loop, whose backbone dipolar groups and conserved arginine side chain are oriented so as to stabilize negative charge in the active site, is found to be particularly important. However, structural features that are not conserved across the PTPase family make significant contributions in particular cases. The PTPases use dipoles more than positive charges to stabilize negative charge in the active site, and this helps to prevent the destabilization of the general acid by the longer range interactions with positive charges. The use of a low protein dielectric constant in the calculations is essential to obtaining calculated protonation states that are consistent with the catalytic mechanism. Models using a higher dielectric constant bring the calculated cysteine pK(a) values closer to some reported experimental values for the free enzyme, but fail to predict the protonation of the general acid, or the maintenance of the thiolate form of the cysteine residue in the Michaelis complex.