diff --git a/tests/unit/core/test_bf_utils.py b/tests/unit/core/test_bf_utils.py
index fc34f35d9c..10ce1fd09e 100644
--- a/tests/unit/core/test_bf_utils.py
+++ b/tests/unit/core/test_bf_utils.py
@@ -25,7 +25,7 @@ def test_get_standardized_ids_columns():
         "0",
         utils.UNNAMED_COLUMN_ID,
         "duplicate",
-        "duplicate.1",
+        "duplicate_1",
         "with_space",
     ]
     assert idx_ids == []
@@ -37,13 +37,13 @@ def test_get_standardized_ids_indexes():
 
     col_ids, idx_ids = utils.get_standardized_ids(col_labels, idx_labels)
 
-    assert col_ids == ["duplicate.2"]
+    assert col_ids == ["duplicate_2"]
     assert idx_ids == [
         "string",
         "0",
         utils.UNNAMED_INDEX_ID,
         "duplicate",
-        "duplicate.1",
+        "duplicate_1",
         "with_space",
     ]
 
diff --git a/third_party/bigframes_vendored/pandas/io/common.py b/third_party/bigframes_vendored/pandas/io/common.py
index 506984e64d..e186f02b5b 100644
--- a/third_party/bigframes_vendored/pandas/io/common.py
+++ b/third_party/bigframes_vendored/pandas/io/common.py
@@ -13,13 +13,13 @@ def dedup_names(
     """
     Rename column names if duplicates exist.
 
-    Currently the renaming is done by appending a period and an autonumeric,
-    but a custom pattern may be supported in the future.
+    Currently the renaming is done by appending a underscore and an
+    autonumeric, but a custom pattern may be supported in the future.
 
     Examples
     ```
     dedup_names(["x", "y", "x", "x"], is_potential_multiindex=False)
-    ['x', 'y', 'x.1', 'x.2']
+    ['x', 'y', 'x_1', 'x_2']
     ```
     """
     names = list(names)  # so we can index
@@ -34,9 +34,9 @@ def dedup_names(
             if is_potential_multiindex:
                 # for mypy
                 assert isinstance(col, tuple)
-                col = col[:-1] + (f"{col[-1]}.{cur_count}",)
+                col = col[:-1] + (f"{col[-1]}_{cur_count}",)
             else:
-                col = f"{col}.{cur_count}"
+                col = f"{col}_{cur_count}"
             cur_count = counts[col]
 
         names[i] = col