diff --git a/markdown/extensions/md_in_html.py b/markdown/extensions/md_in_html.py
index 3518d059f..174224ab9 100644
--- a/markdown/extensions/md_in_html.py
+++ b/markdown/extensions/md_in_html.py
@@ -17,6 +17,7 @@
 from . import Extension
 from ..blockprocessors import BlockProcessor
 from ..preprocessors import Preprocessor
+from ..postprocessors import RawHtmlPostprocessor
 from .. import util
 from ..htmlparser import HTMLExtractor
 import xml.etree.ElementTree as etree
@@ -263,6 +264,15 @@ def run(self, parent, blocks):
         return False
 
 
+class MarkdownInHTMLPostprocessor(RawHtmlPostprocessor):
+    def stash_to_string(self, text):
+        """ Override default to handle any etree elements still in the stash. """
+        if isinstance(text, etree.Element):
+            return self.md.serializer(text)
+        else:
+            return str(text)
+
+
 class MarkdownInHtmlExtension(Extension):
     """Add Markdown parsing in HTML to Markdown class."""
 
@@ -275,6 +285,8 @@ def extendMarkdown(self, md):
         md.parser.blockprocessors.register(
             MarkdownInHtmlProcessor(md.parser), 'markdown_block', 105
         )
+        # Replace raw HTML postprocessor
+        md.postprocessors.register(MarkdownInHTMLPostprocessor(md), 'raw_html', 30)
 
 
 def makeExtension(**kwargs):  # pragma: no cover
diff --git a/markdown/postprocessors.py b/markdown/postprocessors.py
index cd32687d2..2e68cd9f8 100644
--- a/markdown/postprocessors.py
+++ b/markdown/postprocessors.py
@@ -69,7 +69,7 @@ def run(self, text):
         """ Iterate over html stash and restore html. """
         replacements = OrderedDict()
         for i in range(self.md.htmlStash.html_counter):
-            html = self.md.htmlStash.rawHtmlBlocks[i]
+            html = self.stash_to_string(self.md.htmlStash.rawHtmlBlocks[i])
             if self.isblocklevel(html):
                 replacements["<p>{}</p>".format(
                     self.md.htmlStash.get_placeholder(i))] = html
@@ -95,6 +95,10 @@ def isblocklevel(self, html):
             return self.md.is_block_level(m.group(1))
         return False
 
+    def stash_to_string(self, text):
+        """ Convert a stashed object to a string. """
+        return str(text)
+
 
 class AndSubstitutePostprocessor(Postprocessor):
     """ Restore valid entities """
diff --git a/tests/test_syntax/extensions/test_md_in_html.py b/tests/test_syntax/extensions/test_md_in_html.py
index b68412c8e..433cdd559 100644
--- a/tests/test_syntax/extensions/test_md_in_html.py
+++ b/tests/test_syntax/extensions/test_md_in_html.py
@@ -23,6 +23,21 @@
 from unittest import TestSuite
 from markdown.test_tools import TestCase
 from ..blocks.test_html_blocks import TestHTMLBlocks
+from markdown import Markdown
+from xml.etree.ElementTree import Element
+
+
+class TestMarkdownInHTMLPostProcessor(TestCase):
+    """ Ensure any remaining elements in HTML stash are properly serialized. """
+
+    def test_stash_to_string(self):
+        # There should be no known cases where this actually happens so we need to
+        # forcefully pass an etree Element to the method to ensure proper behavior.
+        element = Element('div')
+        element.text = 'Foo bar.'
+        md = Markdown(extensions=['md_in_html'])
+        result = md.postprocessors['raw_html'].stash_to_string(element)
+        self.assertEqual(result, '<div>Foo bar.</div>')
 
 
 class TestDefaultwMdInHTML(TestHTMLBlocks):
@@ -758,7 +773,7 @@ def test_md1_nested_footnote_ref(self):
 def load_tests(loader, tests, pattern):
     ''' Ensure TestHTMLBlocks doesn't get run twice by excluding it here. '''
     suite = TestSuite()
-    for test_class in [TestDefaultwMdInHTML, TestMdInHTML]:
+    for test_class in [TestDefaultwMdInHTML, TestMdInHTML, TestMarkdownInHTMLPostProcessor]:
         tests = loader.loadTestsFromTestCase(test_class)
         suite.addTests(tests)
     return suite