The method for deriving a stage-discharge relationship has a significant impact on the shape of the river's rating curve. We compare rating curves for a single gauging station on a mutiple-channelled river in Australia compiled using three different methods - the Urban Runoff and Basin Systems (URBS) rainfall-runoff model, an empirically-based velocity-area method, and the predictive Hydrologic Engineering Centre-River Analysis System (HEC-RAS) computer model. The rainfall-runoff model was found to predict lower discharges for stage heights over 3.5 m than both the empirically-based velocity-area method and the HEC-RAS model. The empirically-based velocity-area model predicts similar discharges to the rainfall-runoff model for stage heights less than 3 m but much higher discharges for larger flood events. The HEC-RAS model predicts higher discharges than both other rating curves at all stage heights probably due to under-estimation of the impact of surface roughness on flow velocity. The three models are discussed with particular reference to their use on multiple-channelled rivers.