diff --git a/haystack/dataclasses/document.py b/haystack/dataclasses/document.py
index 7792eaee610..c65c4a7b88e 100644
--- a/haystack/dataclasses/document.py
+++ b/haystack/dataclasses/document.py
@@ -93,11 +93,16 @@ def __eq__(self, other: object) -> bool:
         """
         Compares Documents for equality.
 
-        Two Documents are considered equals if their dictionary representation is identical.
+        Two Documents are considered equal if their field values, including the `meta` mapping, are
+        identical.
+
+        The unflattened dictionary representation is used so that metadata keys that collide with
+        top-level Document fields (e.g. ``id``, ``content``, ``score``) are preserved during the
+        comparison instead of being silently overwritten by the Document's own fields. See #11969.
         """
         if type(self) != type(other):
             return False
-        return self.to_dict() == other.to_dict()
+        return self.to_dict(flatten=False) == other.to_dict(flatten=False)
 
     def __post_init__(self) -> None:
         """
diff --git a/releasenotes/notes/document-eq-metadata-collision-fix-11969.yaml b/releasenotes/notes/document-eq-metadata-collision-fix-11969.yaml
new file mode 100644
index 00000000000..631d63e5c1b
--- /dev/null
+++ b/releasenotes/notes/document-eq-metadata-collision-fix-11969.yaml
@@ -0,0 +1,8 @@
+---
+fixes:
+  - |
+    Fixed `Document.__eq__` reporting two Documents as equal when their `meta` mapping contained
+    keys that collide with top-level Document fields such as `id`, `content`, or `score`. The
+    equality comparison now uses the unflattened dictionary representation, so metadata keys
+    that share a name with a Document field are preserved during comparison instead of being
+    silently overwritten by the Document's own field values.
diff --git a/test/dataclasses/test_document.py b/test/dataclasses/test_document.py
index 18e6566328c..2c7a151ba9f 100644
--- a/test/dataclasses/test_document.py
+++ b/test/dataclasses/test_document.py
@@ -123,6 +123,41 @@ def test_basic_equality_id():
     assert doc1 != doc2
 
 
+def test_equality_preserves_meta_keys_colliding_with_top_level_fields():
+    """
+    Regression test for #11969.
+
+    Equality must not be reported for two Documents whose `meta` contains keys that collide with
+    top-level Document fields (``id``, ``content``, ``score``). Previously ``Document.__eq__``
+    compared the flattened ``to_dict()`` output, which silently overwrote colliding meta keys
+    with the Document's own field values during comparison.
+    """
+    doc1 = Document(id="same_id", content="hello", meta={"id": "different1"})
+    doc2 = Document(id="same_id", content="hello", meta={"id": "different2"})
+
+    assert doc1 != doc2
+
+    doc3 = Document(content="hello", meta={"score": 0.5})
+    doc4 = Document(content="hello", meta={"score": 0.9})
+    assert doc3 != doc4
+
+    doc5 = Document(content="hello", meta={"content": "from_meta"})
+    doc6 = Document(content="hello", meta={"content": "also_from_meta_but_different"})
+    assert doc5 != doc6
+
+
+def test_equality_with_colliding_meta_keys_when_meta_actually_matches():
+    """
+    Companion to the regression test for #11969.
+
+    When two Documents have the same top-level fields AND the same colliding meta key, they
+    must still be considered equal.
+    """
+    doc1 = Document(id="same_id", content="hello", meta={"id": "x"})
+    doc2 = Document(id="same_id", content="hello", meta={"id": "x"})
+    assert doc1 == doc2
+
+
 def test_id_is_independent_of_meta_key_order():
     doc1 = Document(content="hello", meta={"a": 1, "b": 2})
     doc2 = Document(content="hello", meta={"b": 2, "a": 1})