The first semantic roles corpus in Persian language, containing about 30,000 sentences from contemporary Persian language, is manually annotated. This corpus, based on the concept of thematic roles of Fillmore, adds a layer of predicate-argument information to the syntactic structures of Persian Dependency Treebank. In this corpus, the verbs, propositional nouns and adjectives are regarded as the predicates of the sentences and are annotated according to their argument structure. The data was prepared based on Conference on Natural Language Learning (CoNLL) dependency format. Semantic tags used as the semantic annotations include thematic roles and functional tags. Thematic roles labels present the argument structure of the predicates of the sentences, and functional tags modify the verb or the whole sentence. The number of thematic roles tags and functional tags are 27 and 15, respectively. The two tags of NEGATION and MODALS are used as the functional tags.