Identifying like rows across tables in a less brute force way.

Question

I have 2 tables. Neither table has a meaningful unique key as they are. I want to create another table by selecting from the 2 tables where all of the fields are the same. I also want to consider null-null to be a match as well as value-null and null-value to be a match for each field.
What i've done is write a function that compares the values and returns a 0 or 1 based on the rules
value, value -> 1
null,null -> 1
value, null -> 1
null, value -> 1
value, different value -> 0
i then call the function during my insert
A simple version would be
insert into Target_Table( Select T1.x,T1.y,T1.z from T1,T2 where
Func(T1.x,T2.x) = 1 and
Func(T1.y,T2.y) = 1 and
Func(T1.z,T2.z) = 1)
It seems to work but only with small data. I believe that i am basically comparing one row to the entire other table behind the scenes. Obviously this breaks down even when the tables are over a few thousand rows.

Answer 1

No. The 2 tables actuially contain mostly the same data but are from 2 different dates. What i'm trying to do is assign a unique key to each record in the results table. After that i'll use it as my "old" table going forward and always take the key from it and assign new keys to any records that show up in the "new" data.
The problem is that some of the fields might get populated between the time each table is created. There is also a possability that they may change, which breaks my method but i'll jump over that hurdle when i get there.