First off, you'd need a unique index so that you can compare like records, otherwise you have no idea what records in table A to compare to table B.
Assuming you have that index :
scan table B for records that don't exist in table A
SET RELATION TO <keyfield> INTO A IN B
SCAN FOR EOF('A')
** This is a record in B that isn't in A
ENDSCAN
scan table A for records that don't exist in table B
For records with the same key, SCATTER to objects and compare for differences.
SET RELATION TO <keyfield> INTO B IN A
SCAN
IF EOF('B')
** This is a record in A that isn't in B
ELSE
** This record exists in B, Now check for equality
SCATTER MEMO NAME oRecordA
SELECT B
SCATTER MEMO NAME oRecordB
IF !COMPOBJ(oRecordA, oRecordB)
** This record is different between A & B
ENDIF
ENDIF
ENDSCAN
Note: The COMPOBJ returns .F. if the fields are defined differently between the two tables.