In deeply incised rivers, bankfull discharge (i.e. the flow filling the channel to the top of the banks) does not represent channel forming flow and increasingly large flows are associated with increasingly large boundary shear stress. In such rivers, solid bank revetments (rip-rap, gabions, retaining wall) are usually constructed to the top of the banks—similarly as in vertically stable rivers—despite the fact that the upper parts of the banks may never be flooded. To optimize the height of solid bank revetments in deeply incised channels, it is thus important to determine whether a flood magnitude can be identified, for which the combination of flow duration and bedload transport rate results in the highest river efficiency to transport bedload and perform geomorphic work. This question was explored in the Morávka River, Czech Carpathians, which deeply incised into non-resistant flysch bedrock over the past few decades. Observations of high-water marks (e.g. trash lines, wash lines) after a flood in 2014 enabled reconstruction of the peak flood stage in the deeply incised reach and the adjacent, vertically stable reach. These observations, together with post-flood measurements of cross-sectional channel geometry, distances between consecutive cross sections and estimates of channel roughness, were used in one-dimensional hydraulic modelling aimed to determine a peak discharge of the flood in a number of cross sections in both reaches. Despite the close proximity of both reaches, markedly higher discharge values were obtained for the incised reach and the discrepancy was used to calibrate roughness coefficients for the incised reach. A flow-duration curve determined on the basis of a 25-year series of daily discharges in the upstream gauging station together with data about channel geometry and roughness in the incised cross sections were used to simulate bedload transport at successive discharges with the BAGS sediment transport model. The calculations indicated the effective discharge for bedload transport in the incised reach to have the recurrence interval of 7–12 years. The upper limit of the reach-average effective discharge is associated with the stage slightly higher than a half of the bankfull depth. This highlights the disparity between effective and bankfull discharges and indicates that in incised channels, channel forming flow should be linked with the former. The vertical extent of the reach-average effective discharge is proposed as an indicator of the optimal, cost-effective height of solid bank-protection structures in the deeply incised channel and the suitability of this approach is discussed.