In the neural machine translation (NMT) paradigm, transformer-based NMT has achieved great progress in recent years. It uses parallel corpus and is based on the stand end-to-end structure. Inspired by the process of translating sentences by translators and the success of templates in other natural language processing tasks, a new method is proposed to incorporate the bilingual translation templates into the Transformer-based NMT. Firstly, the template extraction method is proposed to generate the parallel templates corpus base on the constituency parse tree. Next, given a sentence to be translated, a fuzzy matching method is proposed to calculate the most possible target translation template from the parallel template corpus. Finally, an effective method is proposed to incorporate the bilingual templates into the Transformer-based NMT decoder. Experiment results achieved in three translation tasks show the effectiveness of the proposed approach.