Crusted surfaces can be major sources of mineral dust emission. Quantitative understanding of dust emission from crusted surfaces is limited, because (1) theories on dust emission are not well tested for such surfaces; and (2) modelling is hampered by a lack of input data sufficient to describe the surface conditions. Combining detailed field measurements with physics-based numerical modelling, we present new insights into dust emission from crusted surfaces. Our measurements confirm that crust erodibility and dust-emission intensity can increase or decrease after previous erosion events. To support interpretation of the measurements and to test the applicability of a state-of-the-art parameterisation to simulate dust emission from crusted surfaces, we apply the dust emission scheme of Shao (2004). Saltation flux, which is input to the scheme, is approximated using the parameterisation of Kawamura (1964) and a scaling factor obtained from observations. Limitations of this approach are discussed. Our results show that the dust emission scheme is suitable to estimate dust emission from crusted surfaces if accurate input data and parameters describing the soil-surface condition are provided. The parameters were optimized for each dust event to achieve a best estimate. The variation of the resulting parameter values confirms the observed variability of dust-emission efficiency between the events and provides further evidence that it was caused by variations in crust erodibility. Our study demonstrates that available physics-based dust-emission parameterisations are able to simulate dust emissions under complicated conditions, but also that refined information on the soil-surface conditions are needed as input to the schemes.