diff --git a/CHANGELOG b/CHANGELOG
index e8afc07cb..24985a5cd 100644
--- a/CHANGELOG
+++ b/CHANGELOG
@@ -1,6 +1,13 @@
 CHANGELOG
 =========
 
+Unreleased
+----------
+
+Fixes:
+- Preserve preceding text when case-insensitive word completion lowercases
+  characters into multiple code points.
+
 3.0.53: 2026-07-26
 ------------------
 
diff --git a/src/prompt_toolkit/completion/word_completer.py b/src/prompt_toolkit/completion/word_completer.py
index 563fdadcb..e44f218c8 100644
--- a/src/prompt_toolkit/completion/word_completer.py
+++ b/src/prompt_toolkit/completion/word_completer.py
@@ -70,6 +70,8 @@ def get_completions(
                 WORD=self.WORD, pattern=self.pattern
             )
 
+        # Lowercasing can change the number of characters in the prefix.
+        start_position = -len(word_before_cursor)
         if self.ignore_case:
             word_before_cursor = word_before_cursor.lower()
 
@@ -89,7 +91,7 @@ def word_matches(word: str) -> bool:
                 display_meta = self.meta_dict.get(a, "")
                 yield Completion(
                     text=a,
-                    start_position=-len(word_before_cursor),
+                    start_position=start_position,
                     display=display,
                     display_meta=display_meta,
                 )
diff --git a/tests/test_completion.py b/tests/test_completion.py
index 8b3541af0..e8cf8c9dd 100644
--- a/tests/test_completion.py
+++ b/tests/test_completion.py
@@ -6,6 +6,9 @@
 import tempfile
 from contextlib import contextmanager
 
+import pytest
+
+from prompt_toolkit.buffer import Buffer
 from prompt_toolkit.completion import (
     CompleteEvent,
     FuzzyWordCompleter,
@@ -295,6 +298,31 @@ def test_word_completer_match_middle():
     assert [c.text for c in completions] == ["abc", "abca"]
 
 
+@pytest.mark.parametrize(
+    "word, candidate",
+    [("İ", "İstanbul"), ("İİ", "İİstanbul"), ("I", "Istanbul"), ("i", "istanbul")],
+)
+@pytest.mark.parametrize("ignore_case", [False, True])
+@pytest.mark.parametrize("match_middle", [False, True])
+@pytest.mark.parametrize("WORD", [False, True])
+def test_word_completer_preserves_text_before_case_expanding_word(
+    word, candidate, ignore_case, match_middle, WORD
+):
+    if match_middle:
+        candidate = "city" + candidate
+    completer = WordCompleter(
+        [candidate], ignore_case=ignore_case, match_middle=match_middle, WORD=WORD
+    )
+    buffer = Buffer(document=Document("before " + word))
+    (completion,) = completer.get_completions(buffer.document, CompleteEvent())
+
+    buffer.apply_completion(completion)
+
+    assert buffer.text == "before " + candidate
+    assert buffer.cursor_position == len(buffer.text)
+    assert completion.start_position == -len(word)
+
+
 def test_word_completer_sentence():
     # With sentence=True
     completer = WordCompleter(