Efficient extracting merchandise information is the key technology for e-commerce searching engine. By analyzing web table characters of HTML pages of e-commerce websites, this article proposes the notion of table area locating, and decomposes the merchandise information extraction into three key processes: searching Preparative Core Areas (PCA), locating Core Area (CA) and extracting attribute values from Core-Area, and then design the algorithm of locating Core Area and the algorithm of extracting attributes names and values. We experimented with the new approach on some HTML pages from various e-commerce websites. The results indicate that this approach can locate merchandise information area and extract attributes names and values efficiently, and have better performance of precise and recall.