A series of five experiments was performed in a two-stage compound channel including various roughness conditions and different types of bridge models, namely, single-opening semi-circular arch bridge model (ASOSC), multiple-opening semi-circular arch bridge model (AMOSC), single-opening elliptic arch bridge model (ASOE), and single-opening straight-deck bridge model with and without piers (DECK) including different span widths. The performances of six different methods for computing backwater around bridge waterways were compared using the experimental data carefully taken on many combinations of cases. The results of the energy method, momentum method, WSPRO method, Yarnell's method, USBPR method, and arch bridge method were compared with experimental results. The results showed that energy method was able to simulate more accurately the measured backwater values than the other methods. The backwater differences between the experimental values and computed values by the energy method are generally within -3.2% and 0.8% in terms of flow depth. A simple generalized function for estimating bridge backwater is also proposed.