Laboratory studies of urban flooding often use geometrically distorted scale models due to the multi-scale nature of these specific flows. The possible bias induced by geometric distortion has never been thoroughly investigated with dedicated laboratory experiments. In this study, we combine experimental and computational modeling to systematically assess the influence of the distortion ratio, that is, the ratio of horizontal to vertical scale factors, on upscaled flow depths and discharge partition between streets. Three flow configurations were considered: a street junction, a street bifurcation, and a small synthetic urban district. When the distortion ratio is varied up to a value of about 5, the upscaled flow depths at the model inlets decrease monotonously and the flow discharge in the branch that conveys the largest portion of the flow is greatly enhanced. For equal flow depths at the model outlets and depending on the configuration, the distortion effect induces a variation of the upstream flow depth approximately from ∼4% to ∼17% and a change in outlet discharge partition up to 24 percentage points. For a distortion ratio above 5, both upscaled upstream flow depths and outlet discharge partition tend to stabilize asymptotically. Our study indicates the direction and magnitude of the bias induced by geometric distortion for a broad range of flow cases, which is valuable for offsetting these effects in practical laboratory studies of urban flooding.