Persistent identifiers (PID) have existed for more than 20 years and have become well established as a means for identifying literature and data on the web. They were invented to address the problem of disappearing internet links, also known as “link rot”, which was seen as undermining the emerging digital record of science. A number of PID systems have since been developed, and their utility for the management of the scientific record has been reviewed. Since the initial launch of the Handle System, we have seen many more uses for persistent identification, besides literature and data. A session at the European Geosciences Union General Assembly 2016 was dedicated to ‘20 years of persistent identifiers – where do we go next?’ A number of contributions from this session have since been developed into full papers that form this Data Science Journal Special Collection, with additional solicited contributions. Together, these papers give us an overview of the use of persistent identifiers in research information infrastructures and possible future directions.