Résumé : This dissertation investigates the role of individual and household heterogeneity in shaping carbon footprints, with a particular focus on gender. Climate change mitigation requires not only systemic transformations but also behavioural shifts at the household level. Understanding who cares more about the environment, who is responsible for higher emissions, and who is more likely to reduce them is essential for designing effective and equitable policies. Chapter 1 develops a novel measure of environmental preferences using geo-localised social media data from 54,000 Swedish individuals. It uncovers a gender gap in pro-environmental concern and behaviours and highlights a well-being cost associated with stronger green preferences, while also showing the potential of preference heterogeneity to predict behaviours such as car ownership and support for environmental policies. Chapter 2 turns to household consumption data to analyse dietary carbon footprints in the UK. Using scanner data covering over 70,000 products and 26,000 emission factors, it disentangles heterogeneity in dietary emissions into differences in food quantities and diet structures. It also identifies the household traits linked to reductions in dietary footprints between 2017 and 2022, showing that higher-emission households have been more likely to decrease their footprints from 2017 to 2022, though along different pathways. Chapters 3 and 4 examine the role of gender in overall household carbon footprints. Chapter 3 exploits Belgian budget data before and during COVID-19 to study the differentiated impact of lockdowns, showing that households with male breadwinners experienced larger reductions, primarily through decreases in transport and dining out. Chapter 4 uses a structural bargaining model with UK data to estimate gender differences in emissions propensity, demonstrating that equalizing bargaining power within couples would lower their emissions by about 2%, comparable to the shift from a meat-based to a vegetarian diet.