Comparing very large XML data. In one embodiment, a first and second XML data received for comparison are stored in the form of respective multiple rows in a database. A first and second set of rows (respectively corresponding to the first and second XML data) having the same value corresponding to a key are retrieved from the database and examined to form a comparison result. The same operations may be performed for each value of the key (and also other keys) present in either of the XML data sets.