I frequently have to deal with this issue. The way I do it is to do it in 2 steps. First, I create a make-table query that has the outer joins pointed one way (which will get all records that exist in both tables and those that exist in just one of them). Then I use a 2nd query to append records to this new table that exist only in the other one.
For example, if I want to combine records from Tables A and B in this manner, I will first run a make-table query that has the outer joins set up so that it includes all records in both A and B, and those records only in A. This will create table C. My 2nd query (the append query) selects the records that are only in B by using an outer join between table B and table C, and setting the criteria on all the fields in table C that are part of the Joins to "is Null" (which will cause the query to drop all records that are in both C and B, leaving you with only those records that appear only in C).
This is hard to explain, but I hope this makes sense. I just started a thread trying to get some ideas for how to do this when I'm joining together 5 tables without going through 30 some-odd steps to deal with all the permutations of which tables the records may or may not be in.