Transmission　error　is　one　of　the　most　important　performance　indicators　for　evaluating　harmonic　drives,　and　can　have　crippling　effects　on　positioning　accuracy　and　stability　of　industrial　robots.　However,　most　of　the　existing　error　analysis　methods　focus　on　a　single　factor,　and　do　not　consider　the　uncertainty　of　dynamic　parameters,　leading　to　evident　limitations.　In　the　present　study,　static　transmission　error　(caused　by　manufacturing　and　assembly　error)　and　dynamic　transmission　error　(generated　by　static　transmission　error　and　dynamic　parameters)　of　a　harmonic　drive　are　modeled.　An　interval　method　is　developed　and　used　to　numerically　express　uncertain　dynamic　parameters　of　the　system.　Chebyshev　polynomials　are　used　to　approximate　the　dynamic　differential　equations　of　the　harmonic　drive,　and　then　the　distribution　of　dynamic　transmission　error　and　its　relationship　with　uncertain　parameters　are　discussed　in　detail.　In　addition,　a　global　sensitivity　analysis　is　carried　out　to　intuitively　demonstrate　how　much　impact　each　parameter　has　on　dynamic　transmission　error.　Our　results　suggest　that　the　moment　of　inertia　J(in)　and　the　torsional　stiffness　coefficient　k(1)　have　a　large　influence　on　dynamic　transmission　error.　Finally,　the　proposed　method　is　validated　by　experiment.　The　method　can　be　adopted　to　determine　the　upper　and　lower　bounds　of　dynamic　transmission　error　of　dynamic　systems　under　the　influence　of　uncertain　parameters　and　provides　a　theoretical　basis　for　transmission　error　optimization　and　compensation.