In this article, we highlight factors that make inferences about people’s attributes in social media difficult. One practical question is, “To what extent do people express content that might strongly signal their emotional state on Twitter and Facebook?” Using automated content analysis, we find that (a) only a relatively small proportion of people’s posts appear to contain direct statements of how they feel, and (b) sentiment in these sentences only weakly predicts a self-reported measure of emotion. A second, more novel question addressed in this article is whether people express life domains that are especially personally important (e.g., religion or academic life for college students) on Twitter and Facebook. Since such self-relevant domains are particularly emotionally meaningful, emotions expressed in these domains may provide a strong signal of people’s general emotional well-being. This article explores survey data from a sample of 795 social media users to shed light on the extent emotions and self-relevant domains are expressed on social media. Findings suggest substantial individual differences in the degree people claim to express self-relevant domains.