A DataSet is the most efficient way to write code for generating grids and searching data in memory. At least thats what I have found. GREAT! But it seems to be totally useless when it comes to performance of loading data. I have a 220,000 record recordset in XML and it takes hours just to load this. When dealing with economic data this is a relatively small recordset. How does MS propose that people adopt these functions if the performance is so bad? Am I missing something here? Are there alternative methods for loading xml data into a DataSet?