This paper examines the Korean sentence stress patterns in both canonical and scrambled sentences and shows that nuclear stress falls on the most deeply embedded XP in the base position whether the sentence is scrambled or not. Hence I refute the claim made in Ishihara (2000) that scrambling is an operation that gives rise to the focus set which is otherwise unavailable. Consequently I maintain Saito`s early observation that scrambling is an optional syntactic operation with no semantic effect as far as focus semantics is concerned. Scrambling is not motivated by focus since the nuclear stress doesn`t change in both scrambled/unscrambled sentences, and sentences with different word order don`t give rise to a different focus set. Also I discuss some discourse-pragmatic factors which should be taken into consideration in sentence stress assignment and claim that the sentence stress patterns deviant from the one predicted by the NSR are not really counter-examples to the NSR. Finally I propose the account of the NSR in Korean for both canonical and scrambled sentences in the phase model in Chomsky (2001).