I think it would be greatly helpful if the insert...returning could be seen as a "select" statement and, thus, being usable in the way I have described. I suspect that the insert...returning is actually implemented as an inser plus a select.
You can create a function which does the INSERT RETURNING and returns the result, and use this function in a subsequent INSERT.
You can also write, in plpgsql : FOR a,b,c IN SELECT ... LOOP INSERT INTO table1 (...) VALUES (a,b) RETURNING your_id INTO _var; INSERT INTO table2 (...) VALUES (c , _var ); END LOOP
