Abstract Fluvial dikes (i.e., levees) can be damaged by overtopping flows, leading to breach formation and severe floods. Accurate prediction of the breach geometry and outflow hydrograph is crucial to achieve sound inundation risk assessment and mitigating measures. A set of laboratory experiments were performed to study overtopping induced fluvial dike breaching. The effects of floodplain tailwater on the breach expansion and outflow were investigated. Results are compared to previous experiments under different main channel inflow discharges combined with a free floodplain. The dike breaching process follows three stages: gradual start of overtopping and slow initiation of dike erosion, fast deepening and widening of the breach with highly transient flows, and flow quasi‐stabilization with or without sustained breach expansion. Using the recorded data, a simplified regression model relating the breach stabilization time and the final breach width to the floodplain confinement and inflow Froude number in the main channel is proposed.