Narrative intelligence is an important part of human cognition, especially in sensemaking and communicating with people. Humans draw on a lifetime of relevant experiences to explain stories, to tell stories, and to help choose the most appropriate actions in real-life settings. Manual authoring the required knowledge presents a significant bottleneck in the creation of systems demonstrating narrative intelligence. In this paper, we describe a novel technique for automatically learning script-like narrative knowledge from crowdsourcing. By leveraging human workers’ collective understanding of social and procedural constructs, we can learn a potentially unlimited range of scripts regarding how real-world situations unfold. We present quantitative evaluations of the learned primitive events and the temporal ordering of events, which suggest we can identify orderings between events with high accuracy.