Stability　prediction　is　a　key　step　to　implement　a　real-time　hybrid　simulation　(RTHS)　testing　successfully.　There　are　two　kinds　of　stability　prediction　methods　based　on　continuous　and　discrete　transfer　function.　In　the　family　of　continuous　transfer　function　based　methods,　the　numerical　and　physical　substructures　are　seen　as　continuous　systems　together　with　loading　system.　In　discrete　family,　all　subsystems　in　RTHS　are　regarded　as　discrete　systems.　Actually,　in　a　real　RTHS,　the　numerical　substructure　is　discrete;　the　physical　substructure　is　continuous.　Meanwhile,　the　signal　coordination　is　needed　to　balance　the　sampling　interval　between　numerical　solution　and　physical　loading,　which　is　ignored　in　the　reported　methods.　In　order　to　predict　the　stability　of　RTHS　system　more　accurately,　this　work　develops　a　discrete-continuous　stability　analysis　method　through　the　concept　of　gain　margin,　which　can　consider　the　performance　of　numerical　substructure,　physical　substructure,　loading　system　and　signal　coordination　comprehensively.　And　the　accuracy　of　the　method　is　verified　by　SIMULINK　simulation　and　experimental　testing.　Based　on　shaking　table　and　actuator　RTHS　systems,　the　performance　of　continuous　and　discrete　methods　is　compared　with　the　proposed　method　analytically.　The　results　show　that　the　discrete　method　and　continuous　method　have　slight　influence　on　the　stability　prediction　with　a　small　integration　step　(e.g　1　ms).　However,　with　the　increase　of　integration　step,　compared　with　the　discrete-continuous　method,　the　discrete　method　and　continuous　method　may　overestimate　or　underestimate　the　stability　of　a　real　RTHS　system.