In　order　to　evaluate　the　status　of　trusted　computing　platform,　we　should　assess　its　risk　based　on　testing.　Through　analyzing　the　risk　indicators　and　the　attack　condition,　we　propose　the　trusted　computing　platform　assessment　method　based　on　multi-domains　Bayesian　network　(MDBN).　The　method　divides　the　Bayesian　network　to　many　sub-domains,　which　can　be　evaluated　independently　in　sub-domains　and　increase　risk　assessment　efficiency　between　them.　Assessment　combining　with　platform　attack　condition　will　be　useful　for　risk　manager　to　find　out　its　status　and　trends.　Experiment　proves　this　risk　evaluation　scheme　is　accumulated　and　objective.