Automatic　and　early　drilling　risks　detection　is　a　significant　issue　of　drilling　cost　reduction　and　drilling　efficiency　improvement.　In　this　paper,　considering　the　inherent　nature　of　drilling　process　data,　a　novel　drilling　risks　monitoring　method　which　can　automatically　detect　drilling　risks　in　real　time　was　developed.　The　newly　proposed　method　integrated　wellbore　hydraulics　model　and　streaming-data-driven　model　parameter　inversion　algorithm　to　realize　drilling　risks　detection.　Through　the　in-depth　analysis　of　several　drilling　risks’　common　response　characteristics,　two　drilling　risk　indicators,　i.e.　the　pressure-loss　factor　and　the　flow-rate　factor,　were　defined　firstly.　Then　the　drilling　risks　monitoring　model　was　established　to　model　the　pressure　and　the　flow　rate　responses.　In　the　model,　the　pressure-loss　factor　and　the　flow-rate　factor　were　model　parameters　which　need　to　be　inversed　using　streaming-data.　Finally,　streaming-data-driven　model　parameter　inversion　algorithm　was　adopted　to　estimate　the　pressure-loss　factor　and　the　flow-rate　factor　in　order　to　detect　drilling　risks　in　real　time.　Besides　that,　the　validity　and　reliability　of　the　method　were　verified　by　experiments.　Laboratory　experiments　conducted　on　a　small-scale　test　facility　and　drilling　field　experiments　conducted　in　a　real　well　had　proved　the　good　performance　of　this　newly　proposed　drilling　risks　monitoring　method.　©　2020　Elsevier　B.V.