To evaluate the accuracy of conventional microsphere method for the measurement of regional cerebral blood flow (rCBF) based on the microsphere model with N-isopropyl-p-[123I]iodoamphetamine (123I-IMP) and SPECT, we performed simulation analysis and clinical studies. Although the microsphere method requires early SPECT scan at a few minutes after injection of the tracer for the accurate measurement of rCBF, the conventional microsphere method, which is generally used, requires more delayed SPECT scan with long scan-duration. In the conventional microsphere method, the delayed SPECT image is corrected to the image at a few minutes after injection of the tracer by using the monitored entire brain activity. By the simulation analysis based on the 2-compartment model (influx; K1 and outflux; k2) using the input function and the entire brain activity obtained from eight subjects respectively, it was found that the conventional microsphere method overestimated the rCBF in the practical range of rCBF and Vd (= K1/k2) given in the 2-compartment model. When the values of rCBF and Vd in the 2-compartment model were given at 0.5 (ml/g/min) and 30 (ml/g) respectively, the rate of overestimation of rCBF by the conventional microsphere method was determined to be 17.3 +/- 0.7% (mean +/- S.D.). Also in clinical studies for eight subjects, the conventional microsphere method overestimated the rCBF compared with those evaluated by non-linear least squares fitting (NLLSF) analysis based on the 2-compartment model. Those results agreed well with the simulation analysis, suggesting the validity of the simulation. The rCBF values clinically estimated by the conventional method were, however, significantly correlated with those calculated by NLLSF analysis, and there were not so much difference between the two quantitative rCBF images obtained by the conventional microsphere method and the microsphere method. Therefore, we conclude that the conventional microsphere method is clinically useful in spite of the overestimation of rCBF.