This chapter explains how we identified and coded protest events over a period of sixteen years in thirty European countries. We present the semi-automated approach that combines natural language processing tools for the identification of relevant news documents and then discuss the manual annotation for a precise coding of protest events from multiple sources that publish news content in English. The semi-automated part allows us to deal with a large number of documents identified through keyword searches, namely 5 million documents. The manual annotation, in turn, guarantees that we are able to distinguish different protest forms, actors, and the issues at play. Our endeavor resulted in a dataset of 30,000 unique protest events that we use in this book to study contentious politics during the Great Recession.