Drought can have substantial impacts on natural systems and human societies. Initiating with meteorological droughts, impacts cascade through to hydrological systems, ecosystems and economies, thereby making droughts a critical element in the management of water resources and (agro)ecosystems. Drought risk assessment is a common practice in many sectors such as water resources management, energy planning, or economic development. However to date drought risk assessment has been neglected for planning initial ecosystem rehabilitation such as in post-mining land. Quantification of the drought variables (severity, duration, frequency) with a proper understanding of their uncertainties is crucial to overcome ecosystem rehabilitation challenges of early vegetation establishment and long-term ecosystem resilience. Simple meteorological drought indices, based on rainfall and/or potential evaporation, have been widely used to derive drought recurrence intervals (also known as severity-duration-frequency (SDF) curves) and even to detect the severity and duration of soil moisture droughts. However, it is debatable whether meteorological data alone are capable of characterising soil moisture droughts – a critical indicator of limited water availability for plants. The purpose of this thesis is to quantify drought frequency across a diverse range of environments in eastern Australia, and to critically evaluate the applicability of the approach to ecosystem management. This is achieved by developing SDF curves as a risk assessment tool for ecosystem rehabilitation. Furthermore, the study assessed the performance of meteorological drought indices in determining soil moisture droughts and the uncertainties involved with drought SDF analyses. This was addressed in three major steps using eastern Australia as a broad-scale study area. Firstly, the SDF curves for 11 selected locations across a broad range of agro-climatic environments in eastern Australia were derived using three drought indices: the Standardized Precipitation Index (SPI), the Reconnaissance Drought Index (RDI), and the Standardized Precipitation-Evapotranspiration Index (SPEI). Based on the indices, bivariate distribution functions of drought severity and duration were derived, and drought SDF curves estimated. Secondly, the capabilities of the SPI and the RDI to detect soil moisture droughts were evaluated using a physically based soil water model (Hydrus-1D) as a representative reference scenario for soil water pressure by assessing the indices’ failure rates (FR) and false alarm rates (FAR). Three alternative time-scales were used to define the indices (three, six and twelve months). Finally, the uncertainties associated with the estimation of SDF curves were assessed by testing sensitivity to using: the three alternative timescales; four marginal probability distribution functions for drought severity and duration; two types of copula; and two sub-periods of the available historical time series. This study revealed that SDFs are site-specific for short-term and long-term droughts across eastern Australia. The developed SDF curves demonstrated potential applicability as a risk management tool in ecosystem rehabilitation, and the utility of different indices for this purpose. The performance analysis indicated that SPI performed better than RDI in terms of correlations with soil water pressure across all locations and soil types. The simulated soil water pressure was most sensitive to uncertainty in parameters of the water retention curve in tropical Cairns, where the simulated soil water pressure rose by up to 4% after changing the default parameter values of the water retention curve by 1%. The results of the FR and FAR analysis demonstrate that the performance of the soil water model may be lower than the performance of meteorological drought indices due to uncertainty in the soil water retention curve. The sensitivity analysis showed that uncertainties in SDF curves were higher for more extreme (severe and prolonged) drought events. The selected timescale of the drought indices had the most significant effect on the derived drought SDF curves while the effect of the selected type of copula and drought index (SPI, RDI and SPEI) was almost negligible. In conclusion, drought SDF curves are a critical component of risk analyses related to early stage ecosystem establishment, thereby can be used to support decisions about site-specific rehabilitation strategies and ecological management practices. Further, the research results encourage the use of simple meteorological drought indices over more complex approaches such as physically based soil water modelling to assess drought SDF. Nevertheless, if accurate soil hydraulic parameters are available, soil water modelling provides physically relevant and plant-specific soil water pressure values such as the water content at the permanent wilting point. This plays a critical role, for example, in the selection of suitable plant establishment approaches (e.g., direct seeding, tube stocking) during early stage ecosystem rehabilitation. Therefore, drought SDF curves can be used to evaluate the drought risk to initial ecosystem establishment, but do not quantify fundamental abiotic and site-specific environmental barriers. The research outcomes and the identified management responses may enhance the capacity of responsible authorities and restoration managers to set up effective risk assessment plans for ecosystem rehabilitations.