This paper focused on the relative clause of complex sentences with a corpus-based method, using the resources and functions of existing trilingual parallel corpus and single language corpora in Chinese, English and Spanish. It contrasted the structures of relative clauses from syntactic and semantic perspectives with the purpose of assessing systematically the similarities and differences among these three languages. The research results showed (1) The results of the contrastive analysis in the trilingual parallel corpus demonstrated differences in structure when the meanings of sentences held the same across the three languages. Furthermore, the analysis of a single language corpus enabled us to reveal the frequency of uses of certain relative clause structures. (2) The high-frequency ''Object in main clause + Subject in relative clause'' structure, also an early acquired feature for L2 learners, all demonstrated the order of SVO in both main and relative clauses in Spanish and in English. The low-frequency ''O + 0'' structure, also late acquired form for L2 learners, showed OV order in relative clauses. Considering their corresponding structures in Chinese, it is suggested that the farther the distance between the relative subject pronouns and the object of the main clause, the harder it is for L3 learners to comprehend, and vice versa.