The research of energy gains begins with the study of the amounts of energy that are not consumed. In this sense, any savings from demand response (which consists of curtailing power used or starting on site generation which may or may not be connected in parallel with the grid) represent a major track for energy recovery and storage. Furthermore, the inclusion of energy produced from renewable energy in the grid, which are subject to fluctuations in production are very important (i.e. solar radiation and wind force are uncontrollable), requires adaptation from energy suppliers. The concept of smart grids is then presented as a solution with establishing, at a regional scale, an intelligent grid that interconnects “conventional” and renewable energy to a set of instrumented buildings whose energy management is driven by an aggregator. Within this concept, final users must necessarily be considered, given the fact that they are the core of a system that is intended with the ability to predict the effectiveness of the changes it induces, such as the introduction of demand responses (thereby reducing the peak demand for electricity). This study aims to develop a standardized diagnostic methodology in order to assess the potential flexibility of energy consuming infrastructures of the smart grid demonstrator of Veolia Environnement (Reflexe project, co-financed by ADEME). We considered the flexibility in terms of comfort of the user. This study was carried out in three phases. First, we visited diversified buildings (hostels, office buildings...) located in the south of France. Those visits allowed us to evaluate flexibility and relevance of the diagnosis. Second, we conducted semi-structured interviews with eleven final users in order to identify standards of comfort and enrich the questionnaire that will be implemented in the last phase. Then, final users from two buildings were asked to complete an online survey. The questionnaire was based on a literature review on the models of behaviour predictions (Ajzen, 1991, Kaiser et al. 1999; Bamberg & Moser, 2007), and also on theories of comfort (Amphoux, 1989; Le Goff, 1994). Several energy-related behaviours were measured: reducing the use of air conditioning / heating, turning off the computer/using the sleep mode; switching off the lights. If the perceptual evaluation of the physical environment (temperature, air and sound quality...) seems to predict the global comfort level, the semi-structured interviews leads us to think that other criteria may influence it, such as the feeling of control or social norms. More generally, this study shows that the assessments of levels of comfort may affect the predictive power of their intentions to perform pro-environmental behaviours in their workplaces. We will discuss the results of the study in terms of energy demand response potential and intervention to implement to foster the behaviour change.