diff --git a/doc/source/whatsnew/v3.0.0.rst b/doc/source/whatsnew/v3.0.0.rst
index 448ceffdaa1eb..df566d2621af9 100644
--- a/doc/source/whatsnew/v3.0.0.rst
+++ b/doc/source/whatsnew/v3.0.0.rst
@@ -992,6 +992,7 @@ Numeric
 ^^^^^^^
 - Bug in :func:`api.types.infer_dtype` returning "mixed" for complex and ``pd.NA`` mix (:issue:`61976`)
 - Bug in :func:`api.types.infer_dtype` returning "mixed-integer-float" for float and ``pd.NA`` mix (:issue:`61621`)
+- Bug in :meth:`DataFrame.combine_first` where Int64 and UInt64 integers with absolute value greater than ``2**53`` would lose precision after the operation. (:issue:`60128`)
 - Bug in :meth:`DataFrame.corr` where numerical precision errors resulted in correlations above ``1.0`` (:issue:`61120`)
 - Bug in :meth:`DataFrame.cov` raises a ``TypeError`` instead of returning potentially incorrect results or other errors (:issue:`53115`)
 - Bug in :meth:`DataFrame.quantile` where the column type was not preserved when ``numeric_only=True`` with a list-like ``q`` produced an empty result (:issue:`59035`)
diff --git a/pandas/core/frame.py b/pandas/core/frame.py
index 9c41b82bbbc8e..dee66a80111f8 100644
--- a/pandas/core/frame.py
+++ b/pandas/core/frame.py
@@ -9141,20 +9141,10 @@ def combine_first(self, other: DataFrame) -> DataFrame:
         1  0.0  3.0  1.0
         2  NaN  3.0  1.0
         """
-        from pandas.core.computation import expressions
 
         def combiner(x: Series, y: Series):
-            mask = x.isna()._values
-
-            x_values = x._values
-            y_values = y._values
-
-            # If the column y in other DataFrame is not in first DataFrame,
-            # just return y_values.
-            if y.name not in self.columns:
-                return y_values
-
-            return expressions.where(mask, y_values, x_values)
+            # GH#60128 The combiner is supposed to preserve EA Dtypes.
+            return y if y.name not in self.columns else y.where(x.isna(), x)
 
         if len(other) == 0:
             combined = self.reindex(
diff --git a/pandas/tests/frame/methods/test_combine_first.py b/pandas/tests/frame/methods/test_combine_first.py
index 1e594043510ea..e93684b4dc90f 100644
--- a/pandas/tests/frame/methods/test_combine_first.py
+++ b/pandas/tests/frame/methods/test_combine_first.py
@@ -398,6 +398,21 @@ def test_combine_first_string_dtype_only_na(self, nullable_string_dtype):
         ).set_index(["a", "b"])
         tm.assert_frame_equal(result, expected)
 
+    @pytest.mark.parametrize(
+        "wide_val, dtype",
+        (
+            (1666880195890293744, "UInt64"),
+            (-1666880195890293744, "Int64"),
+        ),
+    )
+    def test_combine_first_preserve_EA_precision(self, wide_val, dtype):
+        # GH#60128
+        df1 = DataFrame({"A": [wide_val, 5]}, dtype=dtype)
+        df2 = DataFrame({"A": [6, 7, wide_val]}, dtype=dtype)
+        result = df1.combine_first(df2)
+        expected = DataFrame({"A": [wide_val, 5, wide_val]}, dtype=dtype)
+        tm.assert_frame_equal(result, expected)
+
 
 @pytest.mark.parametrize(
     "scalar1, scalar2",