Timber is the only widely used construction material we can grow. The wood from which it comeshas evolved to provide structural support for the tree and to act as a conduit for fluid flow. Theseflow paths are crucial for engineers to exploit the full potential of timber, by allowing impregnationwith liquids that modify the properties or resilience of this natural material. Accurately predictingthe transport of these liquids enables more efficient industrial timber treatment processes to be de-veloped, thereby extending the scope to use this sustainable construction material; moreover, it isof fundamental scientific value — as a fluid flow within a natural porous medium. Both structuraland transport properties of wood depend on its micro-structure but, while a substantial body ofresearch relates the structural performance of wood to its detailed architecture, no such knowledgeexists for the transport properties. We present a model, based on increasingly refined geometricparameters, that accurately predicts the time-dependent ingress of liquids within softwood timber,thereby addressing this long-standing scientific challenge. Moreover, we show that for the minimal-istic parameterisation the model predicts ingress with a square-root-of-time behaviour. However,experimental data show a potentially significant departure from this√tbehaviour — a departurewhich is successfully predicted by our more advanced parametrisation. Our parameterisation of thetimber microstructure was informed by computed tomographic measurements; model predictionswere validated by comparison with experimental data. We show that accurate predictions requirestatistical representation of the variability in the timber pore space. The collapse of our dimen-sionless experimental data demonstrates clear potential for our results to be up-scaled to industrialtreatment processes.