Measurement of pH in aqueous solutions at up to 300° and 150–300 bar is reviewed. Potentiometric membrane electrodes are identified as the sensors giving the most immediate hope of being practical. Zirconia membranes work well above 200° and in alkaline solution, whereas glass membranes are best up to 150° and in acidic solutions. Both membranes are largely free from interferences. Metal—metal oxide electrodes offer poor prospects, deviating from the ideal Nernstian response at all temperatures and being susceptible to interference from many redox and complexing agents, but systems based on iridium oxide have some promise. The hydrogen electrode remains the standard for pH measurement, but its analytical application is limited by the need to know the hydrogen partial pressure. A practical solution to this problem has yet to be found, except in restricted and artificial circumstances. Palladium hydride electrodes may be useful up to about 200°, but in hydrogen-saturated waters revert to being hydrogen electrodes in any case. Non-potentiometric pH measurements with semiconducting oxides have been shown to be possible, but there are many unanswered questions about possible interferences. Considerable extra instrumentation is required, compared with potentiometry. Fibre-optic sensors based on indicator dyes have been investigated at room temperature, and have the great merit of not requiring a reference electrode. They seem, however, prone to many interferences and have an inherently limited working range of ∼ 2 pH. No measurements at high temperature have been reported. Improved reference electrodes for potentiometric systems are still needed, although there have been advances in the design of external pressure-compensated electrodes working at room temperature. The silver—silver chloride system is still the one most favoured. There has been little rigorous work on standard buffer solutions at above 100° and none at above 200°. Neutral and alkaline buffers are especially needed. The establishment of proper pH standards for high-temperature work would make the testing of sensors both speedier and more reliable. Doubtless because of the experimental difficulties involved, few measurements have actually been made at high temperature, and those in a rather restricted range of conditions. In particular, measurements in dilute, poorly buffered, solutions, which provide the most rigorous test of a system's capability, are completely lacking.