This dataset consists of 1634 biomedical abstracts, expert-annotated for the purpose of extracting information about the efficacy of drug combinations from the scientific literature. Beyond its practical utility, the dataset also presents a unique NLP challenge, as the first relation extraction dataset consisting of variable-length relations. Furthermore, the relations in this dataset predominantly require language understanding beyond the sentence level, adding to the challenge of this task. We provide a promising baseline model (see the paper/repo) and identify clear areas for further improvement. We ask that new methods on this dataset are posted to our public leaderboard to improve visibility: https://leaderboard.allenai.org/drug_combo/submissions/public