Rangelands support many ecosystem services important to humans, including climate regulation. They also have a significant role to play in the mitigation of greenhouse gases. However, the capacity of any rangeland to do this depends foremost upon the condition of biodiversity, and the functioning of its ecosystems. Considerable research has been undertaken on rangeland condition but it has not yet included the assessment of biodiversity (plants, animals and microbes) as a primary focus. Rangeland managers have struggled to assess biodiversity condition because it is rarely defined, is everywhere (so what do you assess?), is always changing in response to natural and human disturbances (so how do you know when it has changed?) and what amount signals management action. Here we present a framework that addresses these issues, and apply it to select surrogates and indicators that are scientifically defensible in biological and planning terms for assessing biodiversity. An arid Australian rangeland region is used as a case study to develop and apply our approach. We were not able to illustrate interpretation of condition because of the absence of long-term monitoring data in Australian rangelands, but we do provide guiding principles about sampling design and analytical methods for interpretation that use raw data rather than multimetrics. We discovered that different management outcomes expected to be informed from assessing biodiversity condition affected surrogate and indicator choice, and that a number indicators were not robust when assessed on conceptual relevance, measurement qualities, feasibility of implementation and policy and management relevance for four different management outcomes. Our work highlights the importance of stating the expected outcomes of biodiversity condition assessments up front, so that indicators relevant to future management are chosen. It also shows that critical thought on the robustness of indicators is warranted, especially as condition assessments under climate change will require information on the functional traits of species. We conclude by assessing the strengths and weaknesses of our framework in relation to environmental planning.