The Corpus for fine-grained analysis and automatic detection of irony on Twitter was carefully annotated by trained annotators (Master’s students in Linguistics) using a detailed annotation scheme for irony categorization, which describes four labels: ‘ironic by means of a polarity contrast’, ‘situational irony’, ‘other verbal irony’ and ‘not ironic’. The full guidelines are also provided. All annotations were done using the Brat online annotation tool (https://brat.nlplab.org/index.html) and were revised by the responsible researcher of this project. The dataset consists of 4791 instances with an irony label and a tweet ID. The dataset is provided as an MS Excel file and a CSV (UTF-8) file.