diff --git a/README.rst b/README.rst
index e339992b..5ca8a169 100644
--- a/README.rst
+++ b/README.rst
@@ -96,7 +96,6 @@ Install dependencies::
    sudo apt-get install \
       zlib1g-dev \
       libjpeg-dev \
-      libxml2 \
       tesseract-ocr \
       qpdf \
       unpaper \
diff --git a/RELEASE_NOTES.rst b/RELEASE_NOTES.rst
index 33114145..57e98e52 100644
--- a/RELEASE_NOTES.rst
+++ b/RELEASE_NOTES.rst
@@ -47,6 +47,7 @@ Changes
    - MuPDF_ tools
    - shell scripts
    - Java and JHOVE_
+   - libxml2
 
 -  Some new external dependencies are required or optional, compared to v2.x:
 
@@ -66,6 +67,10 @@ Changes
 Release candidates
 ------------------
 
+-  rc6:
+
+   - dropped libxml2 (Python lxml) since Python 3's internal XML parser is sufficient
+
 -  rc5:
 
    - dropped Java and JHOVE in favour of qpdf
diff --git a/ocrmypdf/hocrtransform.py b/ocrmypdf/hocrtransform.py
index 893c728b..6ceaba23 100755
--- a/ocrmypdf/hocrtransform.py
+++ b/ocrmypdf/hocrtransform.py
@@ -9,7 +9,7 @@
 ##############################################################################
 from reportlab.pdfgen.canvas import Canvas
 from reportlab.lib.units import inch
-from lxml import etree as ElementTree
+from xml.etree import ElementTree
 from PIL import Image
 from collections import namedtuple
 import re
@@ -35,8 +35,7 @@ class HocrTransform():
         self.dpi = dpi
         self.boxPattern = re.compile(r'bbox((\s+\d+){4})')
 
-        self.hocr = ElementTree.ElementTree()
-        self.hocr.parse(hocrFileName)
+        self.hocr = ElementTree.parse(hocrFileName)
 
         # if the hOCR file has a namespace, ElementTree requires its use to
         # find elements
diff --git a/setup.py b/setup.py
index babbc79d..cf8086d9 100644
--- a/setup.py
+++ b/setup.py
@@ -203,7 +203,6 @@ setup(
     install_requires=[
         'ruffus>=2.6.3',
         'Pillow>=2.4.0',
-        'lxml>=3.3.3',
         'reportlab>=3.1.44',
         'PyPDF2>=1.25.1'
     ],