Several cerebral oximeters based on near-infrared spectroscopy (NIRS) are commercially available that determine tissue oxygen saturation (StO). One problem is an inconsistency of StO readings between different brands of instruments. Liquid blood phantoms mimicking optical properties of the neonatal head enable quantitative device comparisons. However, occasionally, the reduced scattering coefficient (μ') of these phantoms decreases over time. AIM To investigate whether this decrease in μ' affects the validity of comparison of these devices. StO was measured by several NIRS oximeters simultaneously on a phantom, which exhibited a particularly strong decrease in μ'. We found that a decrease in μ' by ≤16% from baseline led to deviations in StO of ≤3%.