To investigate the effects of changes in biopolymer composition on moisture in acetylated poplar wood (Populus euramericana Cv.), the acetylation of control wood was compared to the acetylation of wood with reduced hemicellulose or lignin content (about 9% reduction of total specimen dry weight in both cases). Time-domain nuclear magnetic resonance relaxometry of water-saturated wood gave spin–spin relaxation times (T 2) of water populations, while deuteration in a sorption balance was used to characterize the hydroxyl accessibility of the wood cell walls. As expected, the acetylation of pyridine-swelled wood reduced hydroxyl accessibility and made the cell wall less accessible to water, resulting in a reduction of cell wall moisture content by about 24% compared with control wood. Hemicellulose loss per se increased the spin–spin relaxation time of cell wall water, while delignification had the opposite effect. The combined effect of hemicellulose removal and acetylation caused more than a 30% decrease of cell wall moisture content when compared with control wood. The acetylated and partially delignified wood cell walls contained higher cell wall moisture content than acetylated wood. An approximate theoretical calculation of hydroxyl accessibility for acetylated wood was in the low range, but it agreed rather well with the measured accessibility, while acetylated and partially hemicellulose-depleted and partially delignified wood for unknown reasons resulted in substantially lower hydroxyl accessibilities than the theoretical estimate.