One of the main bottlenecks for clinical natural language processing is the lack of access to annotated corpora generated by domain experts. Due to the complexity of the medical language and the heavy use of specialized terminologies the construction of labor-intensive Gold Standard datasets within the biomedical and clinical field is of critical importance and practical value. In order to increase the competitive use of manually annotated clinical case studies, a specialized type of medical article that shows the highest degree of similarity to real clinical texts, we have constructed the second Biomedical Abbreviation Recognition and Resolution (BARR2) corpus as part of a shared task posed at IberEval 2018. This corpus contains Spanish clinical case study sections from a variety of clinical disciplines. The BARR2 clinical cases were annotated by hand in an exhaustive manner by domain experts to label not only all abbreviation mentions but also their corresponding definitions. A subset of these definitions was also manually mapped to control vocabulary resources, in particular to SNOMED CT. We have used this corpus as a resource to train and evaluate participating systems of the BARR2 track. Here we summarize the creation of the BARR2 corpus and the annotation process to create the training, testing and development set for participants. More information about this campaign can be found at: http://temu. bsc. es/BARR2.