Cerebral blood flow (CBF) can be quantified non-invasively using the brain perfusion index (BPI), which is determined using radionuclide angiographic data obtained through the use of technetium-99m hexamethylpropylene amine oxime (99mTc-HMPAO). The BPI is generally calculated using graphical analysis (GA). In this study, BPI was measured using spectral analysis (SA), and the usefulness of SA was compared with that of GA. Thirteen patients with various brain diseases and four healthy male volunteers were examined using radionuclide angiography with 99mTc-HMPAO. The BPI was measured for each subject using both SA and GA. In the four healthy volunteers, the BPI was examined at rest and after the intravenous administration of 1 g of acetazolamide (ACZ). An H2(15)O PET examination was also performed in the 13 patients; the BPIS and BPIG values were compared with the CBF measurements obtained using H2(15)O PET (CBFPET). The BPI values obtained by SA (BPIS) (x) and by GA (BPIG) (y) were correlated (y = 0.568x + 0.055, r = 0.901) in the 13 patients and four healthy volunteers at rest, although the BPIG values were underestimated by 36.1 +/- 7.5% (mean +/- SD) compared with the BPIS values. The degree of underestimation tended to increase with increasing BPIS values. The increase in the BPIS was 32.1 +/- 8.0% after the intravenous administration of ACZ, while the increase in BPIG was only 8.1 +/- 2.8%. This discrepancy was considered to be the result of the BPIG values being affected by the first-pass extraction fraction of the tracer. Although both BPIS and BPIG values were significantly correlated with the CBFPET values, the correlation coefficient for BPIS was higher than that for BPIG (BPIS: r = 0.881; BPIG: r = 0.832). These results suggest that SA produces a more reliable BPI for quantifying CBF using 99mTc-HMPAO than the conventional method using GA. The SA method should be especially useful for activation studies involving pharmacological intervention and/or clinical cases with an increased CBF.