Chunk alignment for the bilingual corpus is the base of Example-based Machine Translation. An anchor-based English-Chinese bilingual chunk alignment model and the corresponding algorithm of alignment are presented in this paper. It can effectively overcome the sparse data problem due to the limited size of the bilingual corpus. In this model, the chunk segmentation disambiguation is delayed to the alignment process, and hence the accuracy of chunk segmentation is improved. The experimental results demonstrate the feasibility and viability of this model.