We have examined the feasibility of measuring local brain pH in vivo with 11CO2 and positron emission tomography. In particular, we have addressed two objections that have been raised against this method: the assumed need to estimate local tissue PCO2 and the rapid fixation of 11C in tissue. From a reexamination of the basic theory, we argue that after administration of 11CO2 the time-dependent distribution of 11C between tissue and blood is independent of the distribution of CO2 already in the body, making it unnecessary to estimate local tissue PCO2. Assuming that the blood--brain barrier is impermeable to bicarbonate ions, there will be equal partial pressures of 11CO2 in blood and tissue at equilibrium. To overcome the problem of fixation in the tissue we have developed a kinetic model of the time-dependent distribution of 11C that accounts for regional variations in blood flow, CO2 extraction, pH, and rate of fixation. The values of the model parameters can be estimated from sequential measurements of tissue activity concentration during administration of 11CO2. Tissue pH can then be calculated from one of the parameter values, a measurement of arterial pH, and known constants. Numerical calculations based on the kinetic model with assumed values of the parameters were used to optimize the experimental design. The calculations show that problems with fixation are much less severe with continuous infusion of activity than with bolus administration. During infusion the tissue curve depends strongly on tissue pH but only weakly on the rate of fixation.(ABSTRACT TRUNCATED AT 250 WORDS)