I have two tables. They each have a field called RecNo. One table has only one RecNo and contains transactional type information, userid remarks etcetera. The other table has this field many times. Historically we have built an index on RecNo in both tables and then linked the tables using RecNo. The application user will pull back the data associated with the RecNo from both tables, modify it and repopulate it back on the server. I understand the data integrity associated with deletes and referential integrity. Are there any speed advantages by using a foreign key and an index over just the indexes on there own?