Emission tomography of positron or gamma emitting inert gases allows calculation of regional cerebral blood flow (rCBF) in cross-sectional slices of human brain. An algorithm is presented for rCBF calculations from a sequence of time averaged tomograms using inhaled 133Xe. The approach is designed to avoid loss of information in brain areas with low flow rates. It is based on linearizing and scaling the early isotope distribution picture (recorded from 0 to 2 min) in rCBF units of ml/100 g/min. This is done by calculating the time constant ki for pixels with high count rate using the entire sequence of four 1 min pictures and using a fixed value for the brain:blood partition coefficient, lambda. The method is essentially based on the bolus distribution principle, and it allows the estimation of blood flow in ischemic areas. Application to positron emission tomography is discussed.