The intravenous administration of H2(15)O makes possible the quantitative evaluation of regional changes in cerebral blood flow (CBF) during various neuronal activations in man. Many variations on this technique have been proposed. We have compared bolus and slow administration procedures and short and long scan lengths. For a short scan period this study suggests that the bolus injection procedure (with a scan length of 60-120 seconds) provides better CBF count statistics and less sensitivity to errors in the input function than the slow infusion procedure. For a long scan length (greater than 3 min) the slow infusion procedure is recommended. Continuation of H2(15)O infusion allows prolongation of the scan length, which enables us to obtain sufficient CBF statistics (the statistical error may be negligibly small if a sufficient amount of H2(15)O is infused). Another advantage of the slow infusion method may be related to the fluctuation of CBF during the scan period. If CBF fluctuates during this period the slow infusion procedure is expected to provide an average estimate of CBF, whereas the bolus injection procedure provides an estimate (an average over 10 seconds) of CBF which is more sensitive to short-term variations. The feasibility of using 15O2 inhalation to measure regional CBF has also been investigated. By means of a dynamic H2(15)O scan which measures the regional distribution volume of water and a static C15O emission scan which measures blood volume prior to the 15O2 scan, both CBF and cerebral metabolic rate for oxygen (CMRO2) have been accurately measured by the 15O2 scan for various physiological conditions. The application of this technique in a human study confirmed the uncoupling of CBF and CMRO2 during motor and visual activation.