diff --git a/doc/source/whatsnew/v1.4.0.rst b/doc/source/whatsnew/v1.4.0.rst
index fd7cb6a69d955..12b44a2bcbd5a 100644
--- a/doc/source/whatsnew/v1.4.0.rst
+++ b/doc/source/whatsnew/v1.4.0.rst
@@ -803,6 +803,7 @@ ExtensionArray
 - Avoid raising ``PerformanceWarning`` about fragmented DataFrame when using many columns with an extension dtype (:issue:`44098`)
 - Bug in :class:`IntegerArray` and :class:`FloatingArray` construction incorrectly coercing mismatched NA values (e.g. ``np.timedelta64("NaT")``) to numeric NA (:issue:`44514`)
 - Bug in :meth:`BooleanArray.__eq__` and :meth:`BooleanArray.__ne__` raising ``TypeError`` on comparison with an incompatible type (like a string). This caused :meth:`DataFrame.replace` to sometimes raise a ``TypeError`` if a nullable boolean column was included (:issue:`44499`)
+- Bug in :meth:`Series.where` with ``ExtensionDtype`` when ``other`` is a NA scalar incompatible with the series dtype (e.g. ``NaT`` with a numeric dtype) incorrectly casting to a compatible NA value (:issue:`44697`)
 -
 
 Styler
diff --git a/pandas/core/frame.py b/pandas/core/frame.py
index 12c6eaa86552f..f0e35a8914870 100644
--- a/pandas/core/frame.py
+++ b/pandas/core/frame.py
@@ -10857,7 +10857,7 @@ def interpolate(
     def where(
         self,
         cond,
-        other=np.nan,
+        other=lib.no_default,
         inplace=False,
         axis=None,
         level=None,
diff --git a/pandas/core/generic.py b/pandas/core/generic.py
index 4aff7acc4c6fb..b0fd343084ef6 100644
--- a/pandas/core/generic.py
+++ b/pandas/core/generic.py
@@ -8937,7 +8937,7 @@ def _align_series(
     def _where(
         self,
         cond,
-        other=np.nan,
+        other=lib.no_default,
         inplace=False,
         axis=None,
         level=None,
diff --git a/pandas/core/internals/blocks.py b/pandas/core/internals/blocks.py
index a172cfa49e5b3..f93f386261005 100644
--- a/pandas/core/internals/blocks.py
+++ b/pandas/core/internals/blocks.py
@@ -962,6 +962,9 @@ def putmask(self, mask, new) -> list[Block]:
         mask, noop = validate_putmask(values.T, mask)
         assert not isinstance(new, (ABCIndex, ABCSeries, ABCDataFrame))
 
+        if new is lib.no_default:
+            new = self.fill_value
+
         # if we are passed a scalar None, convert it here
         if not self.is_object and is_valid_na_for_dtype(new, self.dtype):
             new = self.fill_value
@@ -1173,6 +1176,9 @@ def where(self, other, cond) -> list[Block]:
 
         icond, noop = validate_putmask(values, ~cond)
 
+        if other is lib.no_default:
+            other = self.fill_value
+
         if is_valid_na_for_dtype(other, self.dtype) and self.dtype != _dtype_obj:
             other = self.fill_value
 
@@ -1640,13 +1646,8 @@ def where(self, other, cond) -> list[Block]:
         other = self._maybe_squeeze_arg(other)
         cond = self._maybe_squeeze_arg(cond)
 
-        if lib.is_scalar(other) and isna(other):
-            # The default `other` for Series / Frame is np.nan
-            # we want to replace that with the correct NA value
-            # for the type
-            # error: Item "dtype[Any]" of "Union[dtype[Any], ExtensionDtype]" has no
-            # attribute "na_value"
-            other = self.dtype.na_value  # type: ignore[union-attr]
+        if other is lib.no_default:
+            other = self.fill_value
 
         icond, noop = validate_putmask(self.values, ~cond)
         if noop:
@@ -1741,6 +1742,8 @@ def where(self, other, cond) -> list[Block]:
         arr = self.values
 
         cond = extract_bool_array(cond)
+        if other is lib.no_default:
+            other = self.fill_value
 
         try:
             res_values = arr.T._where(cond, other).T
diff --git a/pandas/core/series.py b/pandas/core/series.py
index f7b3d36bb82e7..5e111ccc727fc 100644
--- a/pandas/core/series.py
+++ b/pandas/core/series.py
@@ -5510,7 +5510,7 @@ def interpolate(
     def where(
         self,
         cond,
-        other=np.nan,
+        other=lib.no_default,
         inplace=False,
         axis=None,
         level=None,
diff --git a/pandas/tests/arrays/integer/test_construction.py b/pandas/tests/arrays/integer/test_construction.py
index be9230175bb5d..e5fd4977ec2b8 100644
--- a/pandas/tests/arrays/integer/test_construction.py
+++ b/pandas/tests/arrays/integer/test_construction.py
@@ -133,11 +133,13 @@ def test_to_integer_array_none_is_nan(a, b):
 )
 def test_to_integer_array_error(values):
     # error in converting existing arrays to IntegerArrays
-    msg = (
-        r"(:?.* cannot be converted to an IntegerDtype)"
-        r"|(invalid literal for int\(\) with base 10: .*)"
-        r"|(:?values must be a 1D list-like)"
-        r"|(Cannot pass scalar)"
+    msg = "|".join(
+        [
+            r"cannot be converted to an IntegerDtype",
+            r"invalid literal for int\(\) with base 10:",
+            r"values must be a 1D list-like",
+            r"Cannot pass scalar",
+        ]
     )
     with pytest.raises((ValueError, TypeError), match=msg):
         pd.array(values, dtype="Int64")
diff --git a/pandas/tests/frame/indexing/test_where.py b/pandas/tests/frame/indexing/test_where.py
index 525bf75476fc7..a079a3a7921d7 100644
--- a/pandas/tests/frame/indexing/test_where.py
+++ b/pandas/tests/frame/indexing/test_where.py
@@ -7,6 +7,8 @@
 import numpy as np
 import pytest
 
+from pandas.compat import np_version_under1p19
+
 from pandas.core.dtypes.common import is_scalar
 
 import pandas as pd
@@ -810,6 +812,68 @@ def test_where_columns_casting():
     tm.assert_frame_equal(expected, result)
 
 
+@pytest.mark.parametrize("as_cat", [True, False])
+def test_where_period_invalid_na(frame_or_series, as_cat, request):
+    # GH#44697
+    idx = pd.period_range("2016-01-01", periods=3, freq="D")
+    if as_cat:
+        idx = idx.astype("category")
+    obj = frame_or_series(idx)
+
+    # NA value that we should *not* cast to Period dtype
+    tdnat = pd.NaT.to_numpy("m8[ns]")
+
+    mask = np.array([True, True, False], ndmin=obj.ndim).T
+
+    if as_cat:
+        msg = (
+            r"Cannot setitem on a Categorical with a new category \(NaT\), "
+            "set the categories first"
+        )
+        if np_version_under1p19:
+            mark = pytest.mark.xfail(
+                reason="When evaluating the f-string to generate the exception "
+                "message, numpy somehow ends up trying to cast None to int, so "
+                "ends up raising TypeError but with an unrelated message."
+            )
+            request.node.add_marker(mark)
+    else:
+        msg = "value should be a 'Period'"
+
+    with pytest.raises(TypeError, match=msg):
+        obj.where(mask, tdnat)
+
+    with pytest.raises(TypeError, match=msg):
+        obj.mask(mask, tdnat)
+
+
+def test_where_nullable_invalid_na(frame_or_series, any_numeric_ea_dtype):
+    # GH#44697
+    arr = pd.array([1, 2, 3], dtype=any_numeric_ea_dtype)
+    obj = frame_or_series(arr)
+
+    mask = np.array([True, True, False], ndmin=obj.ndim).T
+
+    msg = "|".join(
+        [
+            r"datetime64\[.{1,2}\] cannot be converted to an? (Integer|Floating)Dtype",
+            r"timedelta64\[.{1,2}\] cannot be converted to an? (Integer|Floating)Dtype",
+            r"int\(\) argument must be a string, a bytes-like object or a number, "
+            "not 'NaTType'",
+            "object cannot be converted to a FloatingDtype",
+            "'values' contains non-numeric NA",
+        ]
+    )
+
+    for null in tm.NP_NAT_OBJECTS + [pd.NaT]:
+        # NaT is an NA value that we should *not* cast to pd.NA dtype
+        with pytest.raises(TypeError, match=msg):
+            obj.where(mask, null)
+
+        with pytest.raises(TypeError, match=msg):
+            obj.mask(mask, null)
+
+
 @given(
     data=st.one_of(
         OPTIONAL_DICTS, OPTIONAL_FLOATS, OPTIONAL_INTS, OPTIONAL_LISTS, OPTIONAL_TEXT