A　current　observer　based　on　an　internal　model　control　is　derived　to　estimate　uncertainties　in　a　permanent　magnet　synchronous　motor　(PMSM)　caused　by　parameter　variations　and　disturbances.　Specially,　a　statespace　model　for　current　loop　is　set　up　as　the　internal　model,　and　an　estimation　law　is　derived　by　robust　Lyapunov　functions　for　the　augmented　state-space　model　to　improve　the　observer　dynamics　and　stability　index.　Superior　control　performance　on　suppressing　current　ripple　for　PMSM　is　verified　in　experiments.