So how unique is the recordID when importing and exporting data between similar files that originate on separate FMS 10 servers?

If I use the calculated field:

myRecordID = Get(CurrentRecordID)

and use it as the matching field when importing with create, am I setting myself up for a big disaster?

Or would it be wiser to use a compound index field using the date/time, user name, and some large random number? Hoping that the multiple servers (7 btw) would not create two duplicate numbers, ever, ever, ever, for millions of records?

Or am I screwed either way?

Maybe one is a more kind screwing?

Get (RecordID) is unique within a table - not even within a file, and certainly not across files. Why don't you use auto-entered serial numbers, with a unique prefix for each source?