The  objectives  of  this  paper  are  to:  propose criteria for testing the applicability of models for rangeland management; and test 2 competing models using those criteria. Predicting change accurately, detecting change, relevance to management and facilitating communication are the performance criteria. Competing models are the range condition model derived from Clement­ sian ecology and the state and transition  model based on  non-equilibrium  ecology. The ability of competing models to predict ecosystem change is assessed in terms of ecosystem organising processes and attributes  such as competition, resistance and resilience, temporal thresholds and feedback. The ability to detect change is viewed in terms of detection at the community and land­ scape scales. Relevance to management and com­ munication are assessed within the context of effectively communicating the effects of manage­ ment decisions on ecosystem processes to both traditional  and non-traditional decision makers and a larger public constituency.  Although  the state and transition model is the most acceptable of the 2 competing models based on the criteria, improvements  are needed. Most notable  is the need for a spatial component in the application of the model for management purposes, and the need for attention to improved communication with a wide range of  resource  users.