The washout of trifluoromethane (CHF3) from rat cerebral cortex was monitored by 19F NMR. After 15 min of inhalation of 67% CHF3/33% O2 the fluorine signal detected was in a steady state. The CHF3 was switched off rapidly at the endotracheal tube and the washout detected with 12-s time resolution. Two models were used to extract flow information, a simple exponential fit and a model which accounts for arterial CHF3 recirculation. In both cases, a two-compartment model fit the data significantly better than a one-compartment model. In both models, the faster time component varied with increasing pCO2, but no significant change in the slow component was detected. At control values of pCO2, there was a small difference in washout rate constants derived from the two models. At high pCO2, when tissue washout was comparable to arterial washout of CHF3, the model which accounted for arterial recirculation gave higher flows. Using this two-compartment model with correction for recirculation, a control flow (pCO2 = 35 mm Hg) of 0.73 +/- 0.04 ml/min/g was measured. Increasing plasma pCO2 increased the apparent flow six- to sevenfold with a 4.4% increase in flow per millimeter of Hg change in CO2. These results are qualitatively in agreement with results found by others using the washout of 133Xe. However, this method yields values for flow that are lower than those obtained using 133Xe washout, probably because of diffusion limitations of CHF3.