<p>Wood exhibits absorption characteristics that deviate from theories that have conventionally been applied to liquid transport in porous building materials, leading to difficulties in interpreting experimental results and determining the essential hygric properties of wood. This study investigated the irregular water absorption patterns of wood, as well as the possible errors and improvements in water absorption testing. The results of long-term water absorption tests using specimens obtained from different wood samples and sealing methods were compared to understand the deviation patterns systematically. In addition, the causes of two typical patterns were discussed based on the water distribution visualized by X-ray computational tomography, absorption tests using non-aqueous liquids, and strain measurements. The results indicated that the use of non-adhesive sealing methods, which were adequate for mineral building materials, caused unexpected local water accumulation near the side and top surfaces, resulting in overestimation of the absorption. In addition, non-negligible increases (by 1.3–3.6 times) in the absorption rates, as measured against the square root of time, were observed during the early stages of absorption. Gradual decreases in the absorption rates were also noted, caused by the heterogeneity of water absorption and unexpected water accumulation close to the top surface. This led to difficulties in determining the capillary absorption coefficient and capillary water content. Approximation methods combining quadratic and linear functions to quantify and compare the initial and later absorption rates were proposed, and adequate test durations were discussed. The findings of this study will help to establish effective test methods, including those for surface treatment and data processing.</p>