Composite indicators are used in ecology to provide a single metric which can assess the state of a system. To fulfil this purpose, they must be constructed with care and rigour to ensure they are depicting intended outcomes. The New South Wales Office of Environment and Heritage constructed two composite indicators to assess condition and pressure for NSW estuaries but these had not been tested to ensure they were giving an accurate picture of estuary health. I evaluated the construction of the composite indicators and created new data-derived models to assist with estuary management. I found that there was no relationship between the pressure and condition composite indicators, despite being based upon a causal-chain framework. The data used to construct the indicators had been strictly constrained in range, which may have obscured patterns in the data. To better elicit those patterns, I constructed univariate and multivariate state-and-transition models. These identified the pressure drivers for each condition indicator independently of the others, and then simultaneously. Where possible, my models were then validated with independent data, although further validation is needed when additional data are available. I concluded that data used in the composite indicators should not be constrained to permit relationships to be observed. The pressure composite indicator should be revised, as I identified potentially redundant component indicators. My univariate models could be used to target action to improve a specific condition indicator, while the multivariate model permits investigation of pressures and condition responses together, enabling action to be directed to improve overall health. In this research, I developed a method for creating and testing composite indicators in estuarine systems that can be applied to other systems. The models are a tool that can be used to assist in decision making and to tailor management actions, improving overall natural resource management.