Near-infrared spectroscopy is a noninvasive bedside technique for moni
toring hemoglobin saturation (HbO2%) in brain vasculature. The method
linearly relates the optical signal detected from the surface of the h
ead to HbO2%. To do so, the method relies on constant transcranial opt
ical path-length and light scattering as well as minimal interference
by tissues overlying the brain. This study examined these premises. Op
tical signals from a dual-wavelength, near-infrared spectrometer were
correlated with sagittal sinus HbO2% in 7 anesthetized piglets subject
ed to 7 different physiological conditions: normoxia, moderate and sev
ere hypoxia, hyperoxia, hypocapnia, hypercapnic hyperoxia, and hypoten
sion. These conditions were induced by varying the inspired O2 concent
ration (7-100%), ventilatory rate (5-35 breaths/min), and blood pressu
re (phlebotomy 20 ml/kg) to force HbO2% over a wide range (5-93%). To
evaluate interference by tissues overlying the brain, correlations wer
e repeated after the scalp and skull were rendered ischemic. Transcran
ial optical pathlength was measured by phase-modulated spectroscopy. L
inear relationships between optical signals and sagittal sinus HbO2% w
ere found with correlation coefficients ranging from -0.89 to -0.99 (p
< 0.05) among animals; however, slope and intercept had coefficients
of variability of approximately 15 and 333%, respectively. Almost iden
tical linear expressions were observed whether scalp and skull were is
chemic or perfused. Transcranial optical pathlength was constant in ea
ch animal, but ranged from 10 to 18 cm among animals. The data indicat
e that the assumptions underlying near infrared spectroscopy are reaso
nably accurate in a given animal, but that the constants for transcran
ial optical pathlength and light scattering are not the same in all an
imals. Since this variability may give rise to errors in quantifying H
bO2% in the clinical setting, further study is recommended before the
method can be used to manage patients.