All very true. The files I use it on are quite small so the script works fine as is. I wont change it till i get a chance/need to test it agian, no point in putting broken code up! ;) But I will certainly update it.

Regarding the keys, the script makes the assumption, or requirment, that your key defines your unique values, so two rows are identical if there keys match, regardless of the other values. This fits my current needs because I'm comparing to reports to make sure they are outputing the same information, but they have different columns, a subset of which should be the unique key. At that point I care about the extra data in the differences because it helps determine what went wrong in the reports. None of this is by way of an excuse, just an explanation of the cause for the script and hopefully helps explain why it does what it does. It would be rather nice to end up with a generic solution that handles all these cases and perhaps I will work that direction. Thanks for the feed back.