Although a great deal of research data should be preserved, either for use/reuse or to validate research results, that is not true of all such data. To determine which research data are valuable sources or resources for research, we have developed a set of practical guidelines in the form of a checklist. It should be emphasised that this checklist is general in nature. It provides a framework for creating selection guidelines for specific disciplines or stakeholders. It summarises the main reasons for selecting research data for long-term preservation. It can be used by individual researchers or research groups, researchers working together in collaboratories, research institutes, university departments, national and international organisations focusing on a specific academic discipline, or funding bodies. It can also be used by managers of data archives, data repositories or heritage institutes. Selection should preferably take place at the time the data are created, if possible in accordance with a data management policy or infrastructure. If that is not possible, selection decisions can also be taken while the data are being entered into a data repository, or at any later time (for example when re-appraising an existing data collection).