<p>The evaluation of dynamical downscaling of Global Climate Models (GCMs) using Regional Climate Models (RCMs) is complicated and challenging because of observational errors and uncertainties. This study assesses the Added Value (AV) for monthly precipitation against multiple observational datasets using the multi-model RCM ensemble outputs for the Coordinated Regional Climate Downscaling Experiment (CORDEX) Australasia domain. These results are further compared against multiple gridded observational datasets to evaluate the impact of observational uncertainties in the evaluation of RCMs and the associated AV. The results show that observational uncertainty plays an important role in the model performance evaluation and, consequently, the AV particularly at local scales. The RCMs produced positive AV over the peak of Australian Alps but not over the steep slopes of Alps largely because of underestimated observed precipitation. Notably, the RCMs show enhanced performance against observational dataset that combines in situ data and satellite-reanalysis estimates, and accounting for precipitation undercatch corrections. Overall, the RCMs consistently shows better performance once the observational uncertainty is included using the Observational Range Adjusted (ORA) statistics. We find that explicitly accounting for the observation uncertainty does not cause substantial changes to the AV at continental scales. However, at local scales the effects of observational uncertainty on the AV can be substantial especially over complex terrain where the observational uncertainty can be large.</p>