diff --git a/docx2everything/converters/markdown_converter.py b/docx2everything/converters/markdown_converter.py
index 0ff408f..475f714 100644
--- a/docx2everything/converters/markdown_converter.py
+++ b/docx2everything/converters/markdown_converter.py
@@ -167,7 +167,8 @@ def parse_paragraph_to_markdown(p_elem, numbering_info=None, hyperlinks=None, im
     # Check for section break
     has_section_break = False
     sectPr = p_elem.find(qn('w:sectPr'))
-    if sectPr is not None:
+    nested_sectPr = pPr.find(qn('w:sectPr')) if pPr is not None else None
+    if sectPr is not None or nested_sectPr is not None:
         has_section_break = True
     
     # Check for heading
diff --git a/tests/test_markdown_section_break.py b/tests/test_markdown_section_break.py
new file mode 100644
index 0000000..ce71a6e
--- /dev/null
+++ b/tests/test_markdown_section_break.py
@@ -0,0 +1,46 @@
+import xml.etree.ElementTree as ET
+
+from docx2everything.converters.markdown_converter import parse_paragraph_to_markdown
+
+
+W_NS = "http://schemas.openxmlformats.org/wordprocessingml/2006/main"
+
+
+def paragraph_with_nested_section_break():
+    xml = f"""
+    <w:p xmlns:w="{W_NS}">
+      <w:pPr>
+        <w:sectPr/>
+      </w:pPr>
+      <w:r><w:t>Body text</w:t></w:r>
+    </w:p>
+    """
+    return ET.fromstring(xml)
+
+
+def paragraph_with_direct_section_break():
+    xml = f"""
+    <w:p xmlns:w="{W_NS}">
+      <w:sectPr/>
+      <w:r><w:t>Body text</w:t></w:r>
+    </w:p>
+    """
+    return ET.fromstring(xml)
+
+
+def test_nested_section_break_outputs_section_break_marker():
+    paragraph = paragraph_with_nested_section_break()
+
+    markdown = parse_paragraph_to_markdown(paragraph)
+
+    assert markdown.startswith("<!-- Section Break -->")
+    assert markdown.endswith("Body text")
+
+
+def test_direct_section_break_still_outputs_section_break_marker():
+    paragraph = paragraph_with_direct_section_break()
+
+    markdown = parse_paragraph_to_markdown(paragraph)
+
+    assert markdown.startswith("<!-- Section Break -->")
+    assert markdown.endswith("Body text")