A method is proposed for measuring brain activity during exercises involving head motion by using functional near-infrared spectroscopy (fNIRS), which investigates cerebral hemodynamics. Obtaining measurements during exercise is difficult because cerebral blood flow changes due to the head motion component (HMC), in addition to neural activity. HMC is an undesirable artifact in the measurement of hemodynamic response caused by neural activity, and as such, it must be estimated and eliminated. In our experiments, cerebral blood flow and head motion were measured during repeated passive forward bending of the subjects. Head motion was measured by 3-D motion capture, and HMC was estimated by deriving a relation between head motion and cerebral blood flow, where the pitch angle was found to be suitable for estimating HMC. In this research, an assumption was made that cerebral blood flow caused by neural activity and that caused by postural change were additive, and thus HMC was eliminated by subtraction.