This　paper　presents　a　multi-scale　stochastic　dynamic　analysis　method　for　offshore　structures.　The　uncertainties　in　the　structural　material　parameters,　such　as　mass　density　and　Young＇s　modulus,　are　considered.　They　are　assumed　to　be　lognormal　distributions　and　represented　by　using　the　Karhunen-Loeve　(KL)　and　Polynomial　Chaos　(PC)　expansions.　Since　the　variance　of　the　output　responses　is　unknown,　the　output　vibration　response　is　represented　by　using　PC　expansion.　The　multi-scale　stochastic　analysis　is　conducted　with　PC　expansions　of　different　orders　representing　responses　at　different　DOFs　defined　as　three　categories,　namely,　important,　less　important　and　the　least　important　ones.　Iterated　Order　Reduced　(IOR)　model　reduction　technique　is　employed　to　remove　the　PC　coefficients　of　slave　DOFs.　Two　numerical　examples　are　taken　to　verify　the　accuracy　and　efficiency　of　the　proposed　method　for　the　multi-scale　stochastic　dynamic　response　analysis　of　offshore　risers.　The　response　statistics　such　as　mean　value　and　variance　can　be　obtained　from　the　proposed　method.　The　results　are　compared　with　those　from　Monte　Carlo　Simulation　(MCS)　and　Stochastic　Finite　Element　Method　(SFEM).　Results　demonstrate　that　the　computational　demand　for　uncertainty　evaluation　is　greatly　reduced,　and　the　accuracy　of　the　results　is　maintained.