Every year since 1946, the General Debate has taken place at the beginning of the United Nations (UN) General Assembly session. Representatives from all UN member states deliver an address, discussing the issues that they consider most important in global politics, revealing their governments’ positions, and seeking to persuade other states of their perspectives. The annual UN General Debate statements provide invaluable information for scholars of international relations – comparable globally and over time. However, these texts are stored as poor quality images without relevant metadata, preventing researchers from applying data science methods. This paper introduces the complete UN General Debate Corpus (UNGDC). Building on a previous incomplete release of UNGDC, we have extended the corpus to cover the entire 1946–present period, included additional data on all speakers and provided advanced search and data visualization tools on a new website. The complete corpus contains over 10,000 speeches from 202 countries, including historical countries – making it the most comprehensive, unique and accessible collection of global political speeches. We discuss the complete UNGDC, provide relevant information for data users and present illustrative examples of how the corpus can be employed to address key questions in world politics.