Many devices and methods are available for the measurement and control of flow rate in open irrigation channels. Among, these devices, sharp crested weirs and sluice gates are commonly used. In order to minimize the disadvantages of weir and gate, they can be combined together in one device yielding a simultaneous flow over the weir and below the gate. In this study the hydraulic characteristics of combined device of weir-gate such as coefficient of discharge and effective parameter on simultaneous discharge were investigated. 38 hydraulic models, divided in five groups were tested on flume with rectangular and circular cross section at the hydraulic laboratory of Isfahan University of Technology. Each model in rectangular flume tested in three conditio a)free flow condition, b) free flow over weir and submerge flow under gate c) submerge flow condition. The result of the experiments indicated that for combined model (in there mentioned flow condition) the coefficient of discharge increases with an increase in dimensionless total head. Three methods for predicting of the combined flow discharge have presented. In the first method, an equal coefficient discharge for weir and gate has assumed and simultaneous flow discharge predicted from combination of available equations for weir and gate. In the second method, a control section over the weir crest has assumed and by using the momentum equations, the flow discharge was found. In the third method, simultaneous flow over combined model has simulated to the flow under a sluice gate. Predictions from these methods have compared with the result of experiments and percentage relative error has used to evaluate the performance of each model. This comparison shows that all methods are able to predict the simultaneous flow over combined model well and first method is the best of them with lowest error. Each model in circular flume tested in the end of flume under free flow condition and simultaneous flow predicted by using first mentioned method. The prediction of proposed model agrees well with observation with a deviation of less than ±5% for about 95% of the data.