>Harry,
>
>>>REPLACE (FIELDNAME) WITH (FIELDGET) FOR &REPCOND
>
>Here and in the SCAN version, do you have an order set to an index tag that is being updated by the replace?

no, the current index field (on which the relationship is of course dependant) is automatically one of the 'protected fields' that can't be amended. And although it might also fail the test (I hadn't thought to test it yet!) it certainly isn't the cause of the failure.

Harry