diff --git a/pandas/core/arrays/categorical.py b/pandas/core/arrays/categorical.py
index ff835eb32f6df..29a172dcdd2c7 100644
--- a/pandas/core/arrays/categorical.py
+++ b/pandas/core/arrays/categorical.py
@@ -373,7 +373,7 @@ def __init__(
         # infer categories in a factorization step further below
 
         if fastpath:
-            self._codes = coerce_indexer_dtype(values, dtype.categories)
+            self._ndarray = coerce_indexer_dtype(values, dtype.categories)
             self._dtype = self._dtype.update_dtype(dtype)
             return
 
@@ -450,7 +450,7 @@ def __init__(
             codes = full_codes
 
         self._dtype = self._dtype.update_dtype(dtype)
-        self._codes = coerce_indexer_dtype(codes, dtype.categories)
+        self._ndarray = coerce_indexer_dtype(codes, dtype.categories)
 
     @property
     def dtype(self) -> CategoricalDtype:
@@ -923,7 +923,7 @@ def set_categories(self, new_categories, ordered=None, rename=False, inplace=Fal
             codes = recode_for_categories(
                 cat.codes, cat.categories, new_dtype.categories
             )
-            cat._codes = codes
+            cat._ndarray = codes
         cat._dtype = new_dtype
 
         if not inplace:
@@ -1096,7 +1096,7 @@ def add_categories(self, new_categories, inplace=False):
 
         cat = self if inplace else self.copy()
         cat._dtype = new_dtype
-        cat._codes = coerce_indexer_dtype(cat._codes, new_dtype.categories)
+        cat._ndarray = coerce_indexer_dtype(cat._ndarray, new_dtype.categories)
         if not inplace:
             return cat
 
@@ -1201,7 +1201,7 @@ def remove_unused_categories(self, inplace=no_default):
             new_categories, ordered=self.ordered
         )
         cat._dtype = new_dtype
-        cat._codes = coerce_indexer_dtype(inv, new_dtype.categories)
+        cat._ndarray = coerce_indexer_dtype(inv, new_dtype.categories)
 
         if not inplace:
             return cat
@@ -1384,6 +1384,10 @@ def __setstate__(self, state):
         if "_dtype" not in state:
             state["_dtype"] = CategoricalDtype(state["_categories"], state["_ordered"])
 
+        if "_codes" in state and "_ndarray" not in state:
+            # backward compat, changed what is property vs attribute
+            state["_ndarray"] = state.pop("_codes")
+
         for k, v in state.items():
             setattr(self, k, v)
 
@@ -1785,11 +1789,11 @@ def fillna(self, value=None, method=None, limit=None):
     # NDArrayBackedExtensionArray compat
 
     @property
-    def _ndarray(self) -> np.ndarray:
-        return self._codes
+    def _codes(self) -> np.ndarray:
+        return self._ndarray
 
     def _from_backing_data(self, arr: np.ndarray) -> Categorical:
-        return self._constructor(arr, dtype=self.dtype, fastpath=True)
+        return type(self)(arr, dtype=self.dtype, fastpath=True)
 
     def _box_func(self, i: int):
         if i == -1:
@@ -1800,7 +1804,7 @@ def _unbox_scalar(self, key) -> int:
         # searchsorted is very performance sensitive. By converting codes
         # to same dtype as self.codes, we get much faster performance.
         code = self.categories.get_loc(key)
-        code = self._codes.dtype.type(code)
+        code = self._ndarray.dtype.type(code)
         return code
 
     # ------------------------------------------------------------------
@@ -2162,7 +2166,7 @@ def unique(self):
         cat = self.copy()
 
         # keep nan in codes
-        cat._codes = unique_codes
+        cat._ndarray = unique_codes
 
         # exclude nan from indexer for categories
         take_codes = unique_codes[unique_codes != -1]
diff --git a/pandas/tests/indexes/categorical/test_category.py b/pandas/tests/indexes/categorical/test_category.py
index 4b2df268f5c1a..8c9caf2e59011 100644
--- a/pandas/tests/indexes/categorical/test_category.py
+++ b/pandas/tests/indexes/categorical/test_category.py
@@ -351,7 +351,7 @@ def test_engine_type(self, dtype, engine_type):
             # having 2**32 - 2**31 categories would be very memory-intensive,
             # so we cheat a bit with the dtype
             ci = CategoricalIndex(range(32768))  # == 2**16 - 2**(16 - 1)
-            ci.values._codes = ci.values._codes.astype("int64")
+            ci.values._ndarray = ci.values._ndarray.astype("int64")
         assert np.issubdtype(ci.codes.dtype, dtype)
         assert isinstance(ci._engine, engine_type)