Textiles are ubiquitous in clothing, medical, sport or hygienic care, but also in engineering or architecture. The control of liquid spreading in these materials is often crucial or desirable, such as e.g., moisture management in functional clothing or durability of fibre reinforced materials. We present a 4D imaging and imaged-based modelling approach to understand the wicking behaviour in textiles made of PET at yarn and fabric scale. In the experimental research, we use fast X-ray tomographic microscopy and neutron radiography to monitor the imbibition process in yarns at micrometre and centimetre scale, respectively. Special attention is given to the liquid transport over the connection of two yarns, called interlaces. At the fabric scale back-light imaging is used to follow the uptake process in fabrics along warp and weft direction. Analysis of the data shows an irregular wicking process characterized by two distinct periods: fast pore filling events followed by long time delays between different pore-to-pore transitions. As a consequence, the water uptake in yarns follows a non-classical behaviour, where the cumulative uptake volume does not follow a square root of time behaviour, as described by Washburn’s law. To model the wicking process, a pore network model is developed based on the typical pore network topology of yarns and waiting times as observed at micron scale. This network model is upscaled to millimetre scale and validated with the neutron measurements, showing good agreement. Finally, the wicking process along war and weft direction is modelled. The pore network approach shows clearly that although the uptake does not follow a square root of time behaviour at pore scale, the uptake process at macroscale becomes more regular due to the interaction of a multitude of pore filling events at the short time. This study provides fundamental insights in the understanding of water uptake in more complex hierarchical porous building materials from pore to macroscale.