The damage triggered by different flood events costs the Italian economy millions of dollars each year. This cost is likely to increase in the future due to climate variability and economic development. In order to avoid or reduce such significant financial losses, risk management requires tools which can provide a reliable estimate of potential flood impacts across the country. Flood loss functions are an internationally accepted method for estimating physical flood damage in urban areas. In this study, we derived a new Flood Loss Function for Italian residential structures (FLF-IT), on the basis of empirical damage data collected from a recent flood event in the region of Emilia–Romagna. The function was developed based on a new Australian approach (FLFA), which represents the confidence limits that exist around the parameterized functional depth–damage relationship. After model calibration, the performance of the model was also validated for the prediction of loss ratios and absolute damage values. In this regard, a three-fold cross-validation procedure was carried out over the empirical sample to measure the range of uncertainty from the actual damage data. The validation procedure shows that the newly derived function performs well (no bias and only 10 % mean absolute error). Results of these validation tests illustrate the importance of model calibration. <br><br> The advantages of the FLF-IT model over other Italian models include calibration with empirical data; consideration of the epistemic uncertainty of data; and the ability to change parameters based on building practices across Italy.