Abstract Assessment of the ecological risk posed to native vegetation from the development of shallow and saline water tables is considered an urgent task in southern Australia. Ecological risk can be defined as the product of the likelihood of an ecological effect and the consequences of that effect. At present, the likelihood of the development of a shallow water table is determined by hydrological modelling at the catchment scale, and this in itself is often equated to ecological risk. In contrast, the ecological consequences of secondary salinity are generally investigated at the patch scale. Translating ecological likelihood and the ecological consequences of risk across these scales has proved problematic, both conceptually and quantitatively. Here we argue that the consideration of ecological risk within the context of the patch- or catchment-scale is based upon human perceptions of spatial units, rather than the ecological scales at which various processes determine vegetation dynamics. By focusing on the processes that determine vegetation dynamics, and the dimensions of these processes, both spatial and temporal scales can be built into conceptual frameworks that aim to understand vegetation change, such as frameworks of alternative stable states. We present an alternative conceptual framework of the ecological risk from secondary salinity based around using the rates of processes, such as salt accumulation in the soil or the frequency of waterlogging, at the scales at which these processes occur. This framework also integrates concepts of vegetation states and transitions between meta-stable states and alternative stable states.