Model validation is one aspect of ecosystem modelling that is not done often. Usually all available data are required to develop a robust model in the first place, and it is rare that models can be revisited years later when additional data have been collected. We revisited an ecosystem response model which was developed for the Coorong, the estuary of the Murray-Darling Basin, Australia, using an approach called ‘ecosystem states’, to assess the model simulations against five years of data that had been collected in the interim. The ecosystem states model has been used extensively to assess competing management actions in the region (e.g. methods of delivering environmental flows) and to simulate the likely future impact of climate change. Since the development of the model in 2008, the additional data has been collected for intervening years, which included both very wet and very dry years. These data enabled an assessment of the model under a range of conditions, including the first data documenting ecological recovery after prolonged drought. This analysis provides an objective assessment of the conditions under which the ecosystem states model performs well and can be reliably used as a tool to assist management, and where additional development is required.