The regional cerebral blood flow can be calculated from the initial slope of the logarithmically displayed clearance curve following intra-arterial injection of Xe-133 (rCBF(initial)). The relationship between this parameter and the values resulting from stochastic (height over area) and compartmental analyses is extensively discussed. Experimental results demonstrate the theoretically expected close relationship between rCBF(initial) and flow of gray substance (rCBF(initial) it, 1 20% to 30% lower than F-g). It is shown how the cerebral clearance curve (normally biexponential) with low flow values becomes gradually monoexponential. Thus only flow of gray substance changes, whereas flow of white substance is independent of CBF infinity, CBF10, was shown to overestimate CBF infinity with about 15% independent of the flow level. Correlation between CBFinitial and CBF10 was linear (r = 0.98) at CBF10 values above 20 ml/100 gm/min. The CBFinitial normal value is found to be 64 +/- 9 ml/100 gm/min, and the interchannel coefficient of variation is 8.2%. A correction for remaining radioactivity from previous measurements is described. Using this, no significant difference was found between repeated resting state measurements. The CBFinitial-Pa-CO2 relationship was found to be best described as exponential, In a group of patients with various intracranial diseases, 1 mm Hg change in Pa-CO2 resulted in 4% change of CBFinitial quite independent of the CBFinitial level.