A fundamental problem in XML data handling is hierarchical data sorting.This paper focuses on this problem and proposes an effective sorting algorithm called XPSort for XML document.XPSort exploits multi-core CPU to parallelize the executions of the mutually independent tasks generated by scanning the XML document;and effectively reduces disk I/Os through data compression

single temporary-file storage and avoidance of tree-matching.XPSort overcomes the issues of inefficient space utilization

large number of random I/Os and inability to handle right-deep tree in NEXSORT

and avoids data redundancy and large disk I/O costs in HERMES.Extensive experiments on XML documents with different characteristics show that XPSort outperforms other existing sorting algorithms.