Summary. Starting from the theory of Hochschild (1983) about emotional labor, workers are supposed to use two different ways to regulate the expressions of emotions required from display rules (Ekman, 1973): deep acting is to modify inner feelings; surface acting is to modify only expressions. Deep acting has good consequences on subjective well-being whereas surface acting has bad consequences on subjective wellbeing. The goal of the research is to understand the way emotional labor (deep and surface acting) affects well-being, comparing two models provided by the literature. Brotheridge & Lee (2002) found authenticity as mediator of this relation, whereas Grandey (2000) suggests that facial feedback could have a role in the relation. So far, both models are coherent with the data. The analysis of the procedures revealed that data were collected only people were supposed to express positive emotions. The models were tested using an experiment in which half of the participants were asked to express positive emotions and the other half negative emotions. Comparing these two conditions and measuring deep acting and surface acting we found evidence supporting the authenticity model rather than the facial feedback model.