Jaime,
It's not too bad a task to just use the DOM. I write a two pass routine that reads the document structure and builds the tables/fields it finds in the document. Then in the second pass it populates the data, along with a PK and FK column that provides the linkage between the tables.
>Thanks for your answer. I see i'll have to use XMLDOM to read the nodes, since this xml file is provided to a web service i wrote by an external company, and i don't think they're using something as flexible as VFP...:-)