Early warning of natural hazards is a topic being increasingly addressed by scientists and demanded by stakeholders responsible for hazard and risk management. In contrast to Early Warning Systems (EWS) for e.g. floods, landslides pose a more challenging problem that is related to the more complex behaviour of the involved processes. In addition, landslide prediction and early warning varies significantly depending on the scale on which they are investigated. In this paper, an overview on existing conceptual approaches for Landslide EWS (LEWS) is provided. The current state of the art and the innovative concepts employed within LEWS are presented for a number of case study systems that have been implemented in different parts of the world. Herein, not only technical characteristics are described but also social aspects such as the integration into general risk management, as well as communication and alerting strategies and risk governance principles are explored.