The aim of this paper is to study the effect of sleep deprivation on the brain dynamics by analysis of Electroencephalogram (EEG) which is the electrical activity of the neuronal population. We recorded the EEG from twelve healthy subjects in the normal states and every 12 hours in the sleep-deprived states. We extracted the information on the brain dynamics from the EEGs using Principal Component Analysis (PCA) and nonlinear analysis. In addition we measured the change in the density of Catecholamine (CA) and Immunoglobulin (IG) in the blood of the subjects to compare the previous biochemical studies on sleep deprivation. With PCA we obtained the 16 spatio-temporal eigenpatterns of the EEG data set. All the subjects have similar eigenpatterns and, especially, the dominant patterns are nearly the same, which means that the dominant pattern has much information on the dynamics of primary brain activity irrespective of individual characteristics. To understand the effect of sleep deprivation on the primary patterns, we compared each component of mean dominant patterns in the normal states with in the sleep-deprived states. We found that the change of component at P4 in the dominant patterns is statistically significant. In the nonlinear analysis we estimated the correlation dimension (D2) and the largest Lyapunov exponent (L1) of the time series of EEGs. It showed that the values of D2 at C3 are reliably increased and those at P4 and O2 are significantly decreased in the 24-hour sleep-deprived states. The values of L1 at P4 is also decreased. In the 36-hour sleep-deprived state the D2s are unchanged in all channels and the L1 at P4 is somewhat increased. It is inferred from these results that the right parietal lobe over the channels of P4, O2 and F4 may be deactivated by sleep deprivation and consequently its function is lowered. In this study we considered the possibility that the EEG analysis with nonlinear methods and PCA may be a helpful tool for investigation into effect of sleep deprivation on the brain dynamics. We suggested from these analyses that the right parietal lobe should be key to the etiology of sleep deprivation leading to lowered information processing.