How are action patterns (re) created in a highly regulated environment? This study aims to present theoretical propositions about how more people experience adaptive routines in a highly regulated environment, but the patterns of action are recomposed and legitimized from the repertoire of practices. Due to its effective role in organizing and its generative power to produce new patterns of action over time (PENTLAND and HAEREM, 2015), routines were applied as units of analysis to understand the constituent elements of the adaptive routine in the social context PENTLAND et al., 2011) of a highly regulated organization, besides allowing the analysis of how the process of (re) creation of action patterns occurs on the basis of Practice-Based Studies (FELDMAN and ORLIKOWSKI, 2011). The regulation was analyzed by addressing the dimensions presented in the work of Pentland (1995), focusing on the organizational structure, culture, technology, coordination and processes, applied to the constructivist model of Orlikowski (2000), which contributed to the construction of the macro view of different Contexts present in the social practices analyzed in the organization. After constructing the macro view of the highly regulated environment context, the Dionysiou and Tsoukas (2013) symbolic interactional generative system model was applied, which provided an understanding of how patterns of action in practice are created, shared, maintained and Altered collectively, by means of collective interaction cycles from the paper-making mechanism of Mead (1934). Thus, through the method of the symbolic interactionist case study, field research was divided into two stages. In the first stage of data collection, a direct observation was performed, which occurred from August 22 to October 14, 2016, generating primary data of the in loco practices, evidenced by field notes, and secondary data Document analysis of the company s internal policies, normative documents, multiple quality management system documents such as work instructions, forms from different sectors and information on IT technologies and systems to carry out company procedures. The second stage of data collection took place through the interviews with 24 people, to capture the actors perception of their practices. The data were systematized by the content analysis (BARDIN, 2011) with the aid of the ATLAS.ti system. The results of the study presented three different ways in which patterns of action may vary in a highly regulated environment, thus maintaining that the type of agencement (GHERARDI, 2016), in other words, the type of connections established in the relationship between agents, agency, (Management Style, Functional Area Type, Accountability, Team Integration, Experience, Cognitive, Motivation, Engagement) Of Personal Resources, Affective Engagement, and Network Amplitude) that all of these aspects together affect the variation of action patterns. Another relevant aspect is the greater the need to include individuals from different groups (other areas and different hierarchical levels) in the collective consensual validation process, the greater the need for interaction cycles for the alignment of actions and, thus, increases The complexity of this process, in addition to reducing its dynamics and fluidity for the adaptation of the actors actions in the face of the uncertainties of the context. Finally, the study proposes weightings for the literature of organizational routines in social practice, exploring the adaptive routines in a highly regulated environment that induce (re) recomposition of action patterns from the repertoire of practices. In this context, the sky is the limit for the creativity of actors to (re) create their action patterns, but it is imperative to keep your feet on the ground revering the mantle of regulation.