The paper argues that ambiguity is not a property of objects but of the relationships of things to classifications and practices. Ambiguity is considered at two levels: firstly, the capacity of things to be subsumed within multiple but equally valid orders, and secondly, the capacity of social actors to articulate and act upon diverse beliefs about the nature of things and their proper relationship to classifications and practices. This opens up the idea of ‘social ontology’: how can we think about and research people's understandings of objects, and how they engage with ambiguity in everyday practices, both as a problem and as a resource and opportunity. Copyright © 2014 John Wiley & Sons, Ltd.