As with bedload transport, wood transport in rivers is governed by several factors such as flow regime, geomorphic configuration of the channel and floodplain, or wood size and shape. Because large-wood tends to be transported during floods, safety and logistical constraints make field measurements difficult. As a result, direct observation and measurements of the conditions of wood transport are scarce. This lack of direct observations and the complexity of the processes involved in wood transport may result in an incomplete understanding of wood transport processes. Numerical modelling provides an alternative approach to addressing some of the unknowns in the dynamics of large-wood in rivers. The aim of this study is to improve the understanding of controls governing wood transport in mountain rivers, combining numerical modelling and direct field observations. By defining different scenarios, we illustrate relationships between the rate of wood transport and discharge, wood size, and river morphology. We test these relationships for a wide, multithread reach and a narrower, partially channelized single-thread reach of the Czarny Dunajec River in the Polish Carpathians. Results indicate that a wide range of quantitative information about wood transport can be obtained from a combination of numerical modelling and field observations and from document contrasting patterns of wood transport in single- and multithread river reaches. On the one hand, log diameter seems to have a greater importance for wood transport in the multithread channel because of shallower flow, lower flow velocity, and lower stream power. Hydrodynamic conditions in the single-thread channel allow transport of large-wood pieces, whereas in the multithread reach, logs with diameters similar to water depth are not being moved. On the other hand, log length also exerts strong control on wood transport, more so in the single-thread than in the multithread reach. In any case, wood transport strongly decreases with increasing piece volume, although this relation is not linear. We also document a nonlinear relationship between wood transport and flood magnitude. A threshold discharge was identified below which wood transport is negligible. This threshold is higher in the multithread reach, while in the single-thread reach floods of lower magnitude are able to transport wood downstream. Wood transport ratio increases with discharge until it reaches an upper threshold or tipping point, and then decreases or increases much more slowly. This threshold is clearly related to bankfull discharge, but it is much higher for the multithread reach than for the single-thread one. Although modelling input and field observations were taken from a specific river, our findings and conclusions are likely to be applicable to a much larger suite of (mountain) rivers.