In order to evaluate the reproducibility of regional cerebral blood flow (rCBF) measurements in pathological brain tissue, serial measurements were carried out in 13 determinations performed in patients who were comatose after severe head injuries within the first 2 weeks after the head trauma. The xenon-133 intra-arterial method was used for the flow measurements, and the flow was studied separately in 16 areas of the brain, producing 197 regional double determinations. The patients were maintained in a steady state during and between the measurements, and only differences of less than 2 mm Hg in the arterial CO2 pressure and less than 5 mm Hg in the perfusion pressure were tolerated. The reproducibility was found to be reasonably good as regards initial slope-flow indices and height/area flow indices and height/area flow values, whereas that of the compartmentally calculated flow parameters was poorer. On dividing the series into severely and moderately pathological areas, it could be shown that the reproducibility of the flor values in the moderately pathological brain areas was acceptable and similar to the results reported by others, whereas the reproducibility in the most severely pathological areas of the brain was rather poor, as regards the flow in the fast compartments. The correlation between the various flow parameters was found to be fairly good.