In this study, we analyzed the language use on Twitter personal exchanges as well as properties of the users’ networks, to study the influence of gender composition on expressions of positive emotions while controlling for the strength of connection between the conversing users. Our findings show that compared to men, women express positive emotions more, especially when interacting with other women. Our findings help the understanding of gender-driven communication patterns in social media, and offer insights for the study of emotion and language. Author Keywords Social media, communication, linguistics, emotion ACM Classification Keywords J.4 [Computer Applications] Social and behavioral sciences. General Terms Theory, Languages