The performance of the Roche tissue pH electrode has been assessed by comparison of values recorded by the the electrode with the pH of arterial blood, in fetal sheep. Observations were made under controlled conditions when the fetal pH was steady, during hypoxia, and after hypoxia. The results showed a highly significant correlation of the values recorded by the electrodes with the pH of arterial blood (r = 0.89, p less than 0.001 during control; and r = 0.86, p less than 0.001 during hypoxia and recovery). However, in about 10% of cases the insertion proved to be unsatisfactory, and in one half of the successful insertions there was a rapid initial drift which lasted up to 45 min. After stabilization, tissue pH values were symmetrically distributed about the atrial pH, with a SD of 0.07 unit. Multiple electrodes in the same fetus gave the same scatter. Movements of the electrode caused significant artefacts. During hypoxia (produced by compression of the cord or administration of gas mixtures low in O2), the electrodes lagged behind the changes in arterial pH by up to 10 min. The conclusion is that the inherent variability of the tissue pH electrode makes it unsuitable as an absolute indicator of fetal well-being, and that it cannot be used alone as an indication for operative intervention. Nevertheless, because of the limitations of conventional techniques, it should be valuable as an adjunct and, in particular, it should help in the interpretation of equivocal fetal heart rate tracings, thereby reducing the risk of fetal death.