The hydrogen bonding of cellulose I structure will be revisited using perdeuterated samples. Previously we proposed a hydrogen bond disorder in native cellulose based on neutron diffraction on samples selectively deuterated on the hydroxyl groups. However, recent calculatoions based on density functional theories predicts the minor hydorgen bonding pattern to have quite high energy. The hydrogen bond pattern should thus be revisited using a better dataset. In our previous study, the sample still contained a large amount of hydrogen (7 out of 10) resulting in an incoherent scattering cross-section 24 times higher than coherent one. We prepared a perdeuterated cellulose sample with similar crystallinity and alignment and expect to improve the data quality by now getting a signal dominated by coherent scattering.