One of the most interesting recent developments in global agri-food systems has been the rapid emergence and elaboration of market audit systems claiming environmental qualities or sustainability. In New Zealand, as a strongly export-oriented, high-value food producer, these environmental market audit systems have emerged as an important pathway for producers to potentially move towards more sustainable production. There have, however, been only sporadic and fractured attempts to study the emerging social practice of sustainable agriculture--particularly in terms of the emergence of new audit disciplines in farming. The ARGOS project in New Zealand was established in 2003 as a longitudinal matched panel study of over 100 farms and orchards using different market audit systems (e.g., organic, integrated or GLOBALG.A.P.). This article reports on the results of social research into the social practice of sustainable agriculture in farm households within the ARGOS projects between 2003 and 2009. Results drawn from multiple social research instruments deployed over six years provide an unparalleled level of empirical data on the social practice of sustainable agriculture under audit disciplines. Using 12 criteria identified in prior literature as contributing a significant social dynamic around sustainable agriculture practices in other contexts, the analysis demonstrated that 9 of these 12 dimensions did demonstrate differences in social practices emerging between (or co-constituting) organic, integrated, or conventional audit disciplines. These differences clustered into three main areas: 1) social and learning/knowledge networks and expertise, 2) key elements of farmer subjectivity--particularly in relation to subjective positioning towards the environment and nature, and 3) the role and importance of environmental dynamics within farm management practices and systems. The findings of the project provide a strong challenge to some older framings of the social practice of sustainable agriculture: particularly those that rely on paradigm-driven evaluation of social motivations, strong determinism of sustainable practice driven by coherent farmer identity, or deploying overly categorical interpretations of what it means to be "organic" or "conventional". The complex patterning of the ARGOS data can only be understood if the social practice of organic, integrated or (even more loosely) conventional production is understood as being co-produced by four dynamics: subjectivity/identity, audit disciplines, industry cultures/structure and time. This reframing of how we might research the social practice of sustainable agriculture opens up important new opportunities for understanding the emergence and impact of new audit disciplines in agriculture. (Contains 2 tables.)