diff --git a/narwhals/_duckdb/dataframe.py b/narwhals/_duckdb/dataframe.py
index 2790d75c2f..33cfc19d2a 100644
--- a/narwhals/_duckdb/dataframe.py
+++ b/narwhals/_duckdb/dataframe.py
@@ -317,9 +317,9 @@ def sort(
         result = self._native_frame.order(
             ",".join(
                 (
-                    f"{col} {desc} nulls last"
+                    f'"{col}" {desc} nulls last'
                     if nulls_last
-                    else f"{col} {desc} nulls first"
+                    else f'"{col}" {desc} nulls first'
                     for col, desc in zip(flat_by, descending_str)
                 )
             )
diff --git a/tests/frame/sort_test.py b/tests/frame/sort_test.py
index 5147c6f562..1ce3414c8e 100644
--- a/tests/frame/sort_test.py
+++ b/tests/frame/sort_test.py
@@ -8,18 +8,18 @@
 
 
 def test_sort(constructor: Constructor) -> None:
-    data = {"a": [1, 3, 2], "b": [4, 4, 6], "z": [7.0, 8, 9]}
+    data = {"an tan": [1, 3, 2], "b": [4, 4, 6], "z": [7.0, 8, 9]}
     df = nw.from_native(constructor(data))
-    result = df.sort("a", "b")
+    result = df.sort("an tan", "b")
     expected = {
-        "a": [1, 2, 3],
+        "an tan": [1, 2, 3],
         "b": [4, 6, 4],
         "z": [7.0, 9.0, 8.0],
     }
     assert_equal_data(result, expected)
-    result = df.sort("a", "b", descending=[True, False])
+    result = df.sort("an tan", "b", descending=[True, False])
     expected = {
-        "a": [3, 2, 1],
+        "an tan": [3, 2, 1],
         "b": [4, 6, 4],
         "z": [8.0, 9.0, 7.0],
     }
@@ -29,14 +29,14 @@ def test_sort(constructor: Constructor) -> None:
 @pytest.mark.parametrize(
     ("nulls_last", "expected"),
     [
-        (True, {"a": [0, 2, 0, -1], "b": [3, 2, 1, None]}),
-        (False, {"a": [-1, 0, 2, 0], "b": [None, 3, 2, 1]}),
+        (True, {"antan desc": [0, 2, 0, -1], "b": [3, 2, 1, None]}),
+        (False, {"antan desc": [-1, 0, 2, 0], "b": [None, 3, 2, 1]}),
     ],
 )
 def test_sort_nulls(
     constructor: Constructor, *, nulls_last: bool, expected: dict[str, float]
 ) -> None:
-    data = {"a": [0, 0, 2, -1], "b": [1, 3, 2, None]}
+    data = {"antan desc": [0, 0, 2, -1], "b": [1, 3, 2, None]}
     df = nw.from_native(constructor(data))
     result = df.sort("b", descending=True, nulls_last=nulls_last)
     assert_equal_data(result, expected)