In many regions, atmospheric rivers are responsible for both a large share of water supply and flood events. Understanding the errors in forecasts of atmospheric rivers is therefore critically important to making confident decisions in the management of water. We have adapted the NCAR Method for Object-Based Diagnostic Evaluation (MODE) software to identify objects that are closely related to atmospheric rivers in model-based analyses and forecasts. This method is applied to the analyses and forecasts created by several models to identify atmospheric river objects making landfall on the US West Coast during winter season 2017. We will demonstrate that this method allows us to measure several forecast performance attributes not commonly reported for operational forecasts of atmospheric rivers. Specifically, we will discuss forecast biases in atmospheric river propagation speed, landfall location, intensity, size, angle of attack and duration of atmospheric river conditions.