There is a plethora of literature linking Extraversion to the experience of positive emotions and Neuroticism to the experience of negative emotions. Further, it has been argued that these relationships have important consequences for well-being. In addition to these main effects of the trait, research on person-situation interactions has shown that individuals have differential reactivity to emotional situations, even identifying some direct causal links using experiments, based on their underlying Extraversion and Neuroticism. However, much less is known about how this differential reactivity might generalize outside of the lab, to naturalistic situations. In Study 1, we test this claim by capitalizing on the natural lockdown that occurred as a result of the COVID-19 pandemic, asking how emotion experience changed due to this lockdown, and whether these changes were moderated by Extraversion, Neuroticism, and the less studied but important trait of Agreeableness. Further, we add an investigation of these potential interactions using facet-level personality. These constructs represent a more specific level of personality analysis than personality traits and have received almost no attention in the literature on the relationships between personality and emotion. However, this increased specificity allows for important clarifying hypotheses about the relationships between personality and emotion to be tested, such as whether the associations between Extraversion and positive emotions are due more to social contact or behavioral activation. In Study 1, we showed that individuals did respond differently to the lockdown based on their underlying Extraversion, Neuroticism and (to a lesser extent) Agreeableness. Further, we showed that the relationship between Extraversion and positive emotions is likely due to behavioral activation more so, or even in place of, social contact. In Study 2, we ask whether these results generalize to a more traditional in lab emotional situation manipulation, using a sad film clip. We capitalize on modern statistical techniques, namely multilevel modelling, to advance the existing work in this area and show that both our results from Study 1, as well as the findings from previous work, which made use of difference scores as dependent variables instead of multilevel models, both generalized well. We discuss the implications of these findings for personality theory, emotion theory, and person by situation interactions, as well as highlight some suggestions for future research.