The　variable　time　delay　and　packet　loss　of　Internet　significantly　degrade　the　manipulability　of　Internet-based　telerobotic　systems　and　even　make　the　systems　unstable.　In　order　to　lessen　the　influence　of　the　variable　time　delay　on　the　system　performance,　a　single-step　time-delay　prediction　algorithm　is　proposed　based　on　the　sparse　multivariate　linear　regression　algorithm　of　variable　sliding　windows.　Then,　a　new　Internet-based　telerobotic　system　is　built　up　in　view　of　event　prediction.　In　the　proposed　system,　a　discrete　path　governor　based　on　time　delay　prediction　is　added　at　the　master　site,　which　generates　a　suitable　event　parameter　online　for　predicting　the　path　by　solving　a　look-ahead　optimization　function,　thus　improving　the　tracking　performance　and　environmental　adaptability　of　the　proposed　system　so　as　to　deal　with　emergent　events.　Moreover,　the　stability　conditions　of　the　proposed　system　are　also　proposed　according　to　the　Lyapunov　stability　theorem.　The　results　of　local　and　trans-regional　experiments　of　a　telerobot　demonstrate　that　the　proposed　methods　can　effectively　solve　the　stability　and　performance　degradation　pro-blems　caused　by　the　variable　time　delay　and　the　packet　loss.