diff --git a/TODO.TXT b/TODO.TXT
index b8d77fe11a..058079770a 100644
--- a/TODO.TXT
+++ b/TODO.TXT
@@ -4,10 +4,6 @@ TODO
 The follow is a list, more or less in the order I intend to do them of things
 that need to be done.  I'm trying to be as granular as possible.
 
-***
-Immediate TODOs:
-Finish refactor of WhereNode so that it just takes connection in as_sql.
-***
 
 2)  Update all old references to ``settings.DATABASE_*`` to reference
     ``settings.DATABASES``.  This includes the following locations
@@ -26,16 +22,8 @@ Finish refactor of WhereNode so that it just takes connection in as_sql.
     * ``dumpdata``: By default dump the ``default`` database.  Later add a
       ``--database`` flag.
 
-    These items will be fixed pending both community consensus, and the API
-    that will go in that's actually necessary for these to happen.
-
     flush, reset, and syncdb need to not prompt the user multiple times.
 
-9)  Fix transaction support.  In practice this means changing all the
-    dictionaries that currently map thread to a boolean to being a dictionary
-    mapping thread to a set of the affected DBs, and changing all the functions
-    that use these dictionaries to handle the changes appropriately.
-
 7)  Remove any references to the global ``django.db.connection`` object in the
     SQL creation process.  This includes(but is probably not limited to):
 
@@ -50,6 +38,7 @@ Finish refactor of WhereNode so that it just takes connection in as_sql.
 
 5)  Add the ``using`` Meta option.  Tests and docs(these are to be assumed at
     each stage from here on out).
+
 8)  Implement some way to create a new ``Query`` for a different backend when
     we switch.  There are several checks against ``self.connection`` prior to
     SQL construction, so we either need to defer all these(which will be
@@ -67,10 +56,5 @@ Finish refactor of WhereNode so that it just takes connection in as_sql.
         * ``DateTimeField.get_db_prep_value``
         * ``DecimalField.get_db_prep_save``
         * ``TimeField.get_db_prep_value``
-    * Implementing the command pattern.
-    * Having the ``QuerySet`` actually store ``Query`` objects for every
-      database in ``settings.DATABASES`` and do all the operations against
-      every single one of them, then when it's time excecute the query just
-      pick the right ``Query`` object to use.  This *does* not scale, though it
-      could probably be done fairly easily.
+
 10) Time permitting add support for a ``DatabaseManager``.
diff --git a/django/db/models/query.py b/django/db/models/query.py
index 2c993cfda5..a4fd172ac2 100644
--- a/django/db/models/query.py
+++ b/django/db/models/query.py
@@ -284,7 +284,7 @@ class QuerySet(object):
         and returning the created object.
         """
         obj = self.model(**kwargs)
-        obj.save(force_insert=True)
+        obj.save(force_insert=True, using=self._using)
         return obj
 
     def get_or_create(self, **kwargs):
diff --git a/django/db/transaction.py b/django/db/transaction.py
index 284636dc15..73c5680f7e 100644
--- a/django/db/transaction.py
+++ b/django/db/transaction.py
@@ -257,8 +257,6 @@ def savepoint_commit(sid, using=None):
 # DECORATORS #
 ##############
 
-# TODO update all of these for multi-db
-
 def autocommit(using_or_func=None):
     """
     Decorator that activates commit on save. This is Django's default behavior;
diff --git a/tests/regressiontests/multiple_database/tests.py b/tests/regressiontests/multiple_database/tests.py
index 3a19469266..f21b873cd3 100644
--- a/tests/regressiontests/multiple_database/tests.py
+++ b/tests/regressiontests/multiple_database/tests.py
@@ -20,17 +20,11 @@ class DatabaseSettingTestCase(TestCase):
 
 class ConnectionTestCase(TestCase):
     def test_queries(self):
-        for connection in connections.all():
-            qn = connection.ops.quote_name
-            cursor = connection.cursor()
-            cursor.execute("""INSERT INTO %(table)s (%(col)s) VALUES (%%s)""" % {
-                'table': qn(Book._meta.db_table),
-                'col': qn(Book._meta.get_field_by_name('title')[0].column),
-            }, ('Dive Into Python',))
+        for db in connections:
+            Book.objects.using(db).create(title="Dive into Python")
 
-        for connection in connections.all():
-            qn = connection.ops.quote_name
-            cursor = connection.cursor()
-            cursor.execute("""SELECT * FROM %(table)s""" % {'table': qn(Book._meta.db_table)})
-            data = cursor.fetchall()
-            self.assertEqual('Dive Into Python', data[0][1])
+        for db in connections:
+            books = Book.objects.all().using(db)
+            self.assertEqual(books.count(), 1)
+            self.assertEqual(len(books), 1)
+            self.assertEqual(books[0].title, "Dive into Python")