diff --git a/test/test_reddit.py b/test/test_reddit.py
new file mode 100644
index 000000000..090ebc965
--- /dev/null
+++ b/test/test_reddit.py
@@ -0,0 +1,70 @@
+from __future__ import unicode_literals
+
+import unittest
+
+from test.helper import FakeYDL
+from youtube_dl.extractor.reddit import RedditRIE
+
+
+class _TestRedditRIE(RedditRIE):
+    def __init__(self, downloader):
+        super(_TestRedditRIE, self).__init__(downloader)
+        self.cookies = {}
+        self.requests = []
+        self.download_kwargs = None
+        self.skip_old_session_cookie = False
+
+    def _get_cookies(self, url):
+        return self.cookies
+
+    def _request_webpage(self, url_or_request, video_id, note=None, errnote=None,
+                         fatal=True, data=None, headers={}, query={}, expected_status=None):
+        self.requests.append((url_or_request, query))
+        if 'old.reddit.com' in url_or_request and not self.skip_old_session_cookie:
+            self.cookies['loid'] = object()
+        elif '/svc/shreddit/' in url_or_request:
+            self.cookies['loid'] = object()
+        return True
+
+    def _download_json(self, url, video_id, **kwargs):
+        self.download_kwargs = kwargs
+        return [{
+            'data': {
+                'children': [{
+                    'data': {
+                        'url': 'https://v.redd.it/test-video',
+                    },
+                }],
+            },
+        }]
+
+    def _extract_m3u8_formats(self, *args, **kwargs):
+        return []
+
+    def _extract_mpd_formats(self, *args, **kwargs):
+        return []
+
+    def _sort_formats(self, formats):
+        pass
+
+
+class TestRedditSession(unittest.TestCase):
+    def setUp(self):
+        self.ie = _TestRedditRIE(FakeYDL())
+
+    def test_initializes_anonymous_session(self):
+        result = self.ie._real_extract(
+            'https://www.reddit.com/r/videos/comments/abc123/example-post/')
+
+        self.assertEqual(result['url'], 'https://v.redd.it/test-video')
+        self.assertEqual(self.ie.requests[0][0], 'https://old.reddit.com/')
+        self.assertEqual(self.ie.download_kwargs.get('expected_status'), 403)
+
+    def test_falls_back_to_shreddit_when_old_reddit_does_not_set_loid(self):
+        self.ie.skip_old_session_cookie = True
+        self.ie._real_extract(
+            'https://www.reddit.com/r/videos/comments/abc123/example-post/')
+
+        self.assertEqual(self.ie.requests[0][0], 'https://old.reddit.com/')
+        self.assertIn('/svc/shreddit/example-post', self.ie.requests[1][0])
+        self.assertEqual(self.ie.requests[1][1]['render-mode'], 'partial')
diff --git a/youtube_dl/extractor/reddit.py b/youtube_dl/extractor/reddit.py
index 222fa0172..25c6c2ccb 100644
--- a/youtube_dl/extractor/reddit.py
+++ b/youtube_dl/extractor/reddit.py
@@ -50,7 +50,7 @@ class RedditIE(InfoExtractor):
 
 
 class RedditRIE(InfoExtractor):
-    _VALID_URL = r'(?P<url>https?://(?:[^/]+\.)?reddit\.com/r/[^/]+/comments/(?P<id>[^/?#&]+))'
+    _VALID_URL = r'(?P<url>https?://(?:[^/]+\.)?reddit\.com/r/[^/]+/comments/(?P<id>[^/?#&]+)(?:/(?P<slug>[^/?#&]+))?)'
     _TESTS = [{
         'url': 'https://www.reddit.com/r/videos/comments/6rrwyj/that_small_heart_attack/',
         'info_dict': {
@@ -97,14 +97,44 @@ class RedditRIE(InfoExtractor):
         'only_matching': True,
     }]
 
+    @property
+    def _is_logged_in(self):
+        return bool(self._get_cookies('https://www.reddit.com/').get('reddit_session'))
+
+    def _real_initialize(self):
+        if not self._is_logged_in:
+            # We need a 'loid' cookie to access the API anonymously.
+            self._request_webpage(
+                'https://old.reddit.com/', None,
+                'Setting up session via old reddit', 'Session request failed', fatal=False)
+
     def _real_extract(self, url):
         mobj = re.match(self._VALID_URL, url)
-        url, video_id = mobj.group('url', 'id')
+        url, video_id, slug = mobj.group('url', 'id', 'slug')
 
         video_id = self._match_id(url)
 
-        data = self._download_json(
-            url + '/.json', video_id)[0]['data']['children'][0]['data']
+        # Fallback for if old reddit session request failed.
+        if not self._is_logged_in and not self._get_cookies('https://www.reddit.com/').get('loid'):
+            self._request_webpage(
+                'https://www.reddit.com/svc/shreddit/%s' % (slug or video_id), video_id,
+                'Setting up session via shreddit', 'Session request failed', fatal=False,
+                query={
+                    'seeker-session': 'false',
+                    'render-mode': 'partial',
+                    'referer': url,
+                })
+
+        try:
+            data = self._download_json(
+                url + '/.json', video_id, expected_status=403)[0]['data']['children'][0]['data']
+        except ExtractorError as e:
+            if isinstance(getattr(e, 'cause', None), ValueError):
+                if self._is_logged_in:
+                    raise ExtractorError(
+                        'Your IP address is unable to access the Reddit API', expected=True)
+                self.raise_login_required('Account authentication is required')
+            raise
 
         video_url = data['url']