Sorry, I missed that part about the buffering. You are right, the records will only be locked by the TableUpdate() command, so there should be no performance difference between the two commands.

In that case, it boils down to personnal preference. UPDATE-SQL being closer to SQL standard and REPLACE closer to XBase. It depends on wich side you are.

---

>This is a fine answer but I'm still wondering, if I use buffering mode 5 (table buffering) I never tough "replace" locks the table, I tough it would be locked only when I say "tableupdate(....). And I tough VFP would return an error message if the record(s) had been changed meanwhile. Am I all wrong?
>
>I'm still confuse with locking when and ow it goes. ((I'm much more confortable when I use MSSQL database as the backend but...)