Abstract This paper investigates the impact of uncertainty in the definition of bathymetry on the performance of estuarine shallow water models. A depth-averaged hydrodynamic model is used to simulate tidal flows in a meso-tidal estuary. Simulations are performed for a range of bathymetric and bed roughness scenarios, which encompass the uncertainties typically encountered in the specification of estuarine and shallow coastal models. The relative sensitivity of model output to uncertainty in bathymetry and bed friction is analysed using Monte Carlo methods, and a new generalised error model is presented that allows various sources of error in bathymetry to be treated as a calibration parameter. Numerical results are compared with observed current speed and water level time series at multiple locations. It is shown that calibration that incorporates bathymetric uncertainty can be significantly more efficient than a classic calibration based only on adjustment of a bed friction coefficient. These findings have wide-ranging implications for the calibration and validation of estuarine and shallow coastal hydrodynamic models. Importantly, our proposed bathymetry calibration framework offers the prospect of significantly improved performance from the current generation of numerical models.