Over the last few years, social media have become part of the daily life of many people, leading scientists to study their users and the data they produce in numerous contexts. For instance, geo-enabled social media provide us with the means to study the dynamics and features of large geographical areas. In this thesis, our goal is to leverage social media to study cities, and their usage by people of different origin (e.g. citizens vs. tourists) and demographics. We design and implement a system that uses Twitter and Instagram as data sources, defining and extracting several features about the city and its users, such as finding points of interests, paths, differentiating users in gender, age, and their role in the city. We also build a proof of concept visualization tool that allows non-scientific users to analyze a city using our extracted data. The system is used for an in-depth analysis, where we compare the usage, as observed through the lens of social media, of cities like Amsterdam, London, Paris, and Rome over a three week period on both Twitter and Instagram. We show that, through social media, it is possible to observe differences in usage patterns, both in the temporal sense, but also in regards to the places that are visited in the city.