When inserting new records via the data loader to an org unrelated to the org from which the data was originally exported, I receive errors concerning the id references that each record contains within the original org (cross reference id error).  From my understanding, this is caused due to the fact that the records that are referenced do not (yet) exist in the destination org.

A specific example:  In the destination org, the desired result is for all records (Accounts, for example) retain their original record owner; the users will also be transferred over.  Also, I would like to ensure that contacts remain related to their respective accounts and so on and so forth for other objects.  I have several questions that I need to find answers for:

1.  What is the best practice for dealing with these id issues?  I have conceived of several plausible solutions, but I am unsure which are most efficient.  a) Insert all of the account records and allow them to default ownership to me (the admin) by leaving the owner field unmapped during loading; then, mass transfer to the appropriate users b)  Import the users first and then manicure the CSV file to ensure that the proper ids are present for each account (using the find and replace all features in excel).  Both of these conceived plans seem somewhat tedious, though.  Is there an easier way?

2.  Is there some sort of method centering around a field that I can add in that would solve these problems?  For example, "Old record owner".

3.  Would someone outline the steps taken while using the data loader?  I'm wondering if I need the command line, for example, or if I'm mapping the fields incorrectly to cause these errors.  Lastly, should I be using some other function of the data loader such as upsert?

Thanks in advance for any help.