Taking the residential buildings within a small community in Xi’an as the study object, a high-precision urban rain-flood model was constructed. The model’s reliability was validated using measured data from monitoring instances of urban inundation during heavy rainfall. The dynamic flooding process inside the residential buildings under six different major rainfall return period design scenarios was simulated. Considering the housing configuration of middle-income families in Xi’an City as the computational unit, the depth thresholds for the flooding of various indoor properties were analyzed, and the relationship curve between the depth of indoor property flooding and the corresponding loss rate was derived. By combining this curve with the flooding depths of different residential buildings in the study area under various major rainfall return periods, accurate data on property losses due to flooding were obtained. The results indicate that under different rainfall recurrence intervals, the inundation depth of the ground floor of the same residential building increases with the larger recurrence interval. For the same recurrence interval, the flooding process and maximum inundation depth of houses in different residential buildings vary due to differences in elevation.