The research on Mongolian Sentimental Vocabulary for natural language processing is in its infancy. However, previous studies in lexicology and rhetoric have rich theoretical foundations. This paper elaborates the concrete methods and implementation procedures of developing a sentiment resource-base by making use of the existing research results of literary corpus and the Mongolian nouns, adjectives and verbs, which include the semantic classification system of Mongolian word that has been developed by the research group, constructing “Mongolian Sentiment Word Resource-base” by matching the “Mongolian Semantic Information Dictionary” database with the labeled corpus. And the paper statistically analyzes the sentimental vocabulary’s distribution characteristics in different texts and different parts of speech, and finally realize the sentimental judgment experiment of Mongolian sentences, which will provide basic resources and methodological references for deeper research on text sentiment analysis.