The presence of a bridge spanning over a river and its floodplain alters the condition of flow. In subcritical conditions, which are typical in alluvial rivers, this flow alteration results in the so-called backwater effect, that is, an increase of the water surface elevation upstream of the structure as a response to increased energy losses. The extension of the expected backwater affected area and the increase in water surface elevation are highly dependent on the river morphology, bridge geometry, flow and floodplain characteristics and may originate upstream flooding. Thus, in many countries, at the stage of design a new bridge (or renovation works), a specific investigation on the backwater effect has to be undertaken to analyse its effect on the flooding of its vicinity. Bridge waterway hydraulics has been widely investigated in terms of practical research and some standard guidelines have been suggested for modelling the backwater effect at bridge crossings. Although the scientific literature has widely proven that hydraulic modelling is affected by many sources of uncertainty, the hydraulic modelling of backwater effects is still undertaken within a deterministic approach, based on the outcomes of a calibrated hydraulic model. This article aims at approaching the prediction of backwater effects at bridge crossings by accounting for the main sources of uncertainty affecting the hydraulic modelling exercise. A bridge over a highly vegetated floodplain (Tallahala Creek, Mississippi, USA) was considered for this analysis, and uncertainty in the model parameters and input data was assessed to predict backwater flood design profiles. Copyright © 2012 John Wiley & Sons, Ltd.