GML documents are extremely verbose since containing large amounts of description information of structure. While some compressors are available to address this problem, no one is designed to support direct querying of the compressed documents so far. In this paper, we propose a new query-supported GML compression tool, called GQComp, based on traditional GML compressors and query-supported XML compressors. GQComp not only achieves high compression ratio as the existing GML compression algorithms, but also implements the attribute query, spatial query and mutual query of the compressed files, through the combination of feature structure tree and spatial index. The experiments indicate that GQComp obtains reasonably good results both at compression ratio and query efficiency.