This paper proposes an integrated analysis method that estimates user intention in utterance by considering the correlative characteristic between each constituent (speech act, concept sequence, and arguments) of the user intention. The proposed method comprises two distinctive steps. In the first step, each constituent of the user intention is analyzed in sequence, the results of which are taken as input for analysis in the second step. In this step, estimation accuracy for user intention can be improved because the correlative characteristic of speech act, concept sequence, and arguments have been considered. Experiment showed that the proposed model improves user intention estimation over the conventional model with error rate reductions in concept sequence, speech act, and arguments by 7.45%p, 14.78%p and 5.84%p respectively.