diff --git a/Lib/test/test_sys.py b/Lib/test/test_sys.py
index 94a09ff549331a..1dc10d8b0a39ac 100644
--- a/Lib/test/test_sys.py
+++ b/Lib/test/test_sys.py
@@ -1539,6 +1539,7 @@ class newstyleclass(object): pass
         samples = ['1'*100, '\xff'*50,
                    '\u0100'*40, '\uffff'*100,
                    '\U00010000'*30, '\U0010ffff'*100]
+        # also update field definitions in test_unicode.test_raiseMemError
         asciifields = "nnb"
         compactfields = asciifields + "nP"
         unicodefields = compactfields + "P"
diff --git a/Lib/test/test_unicode.py b/Lib/test/test_unicode.py
index 64abc0c761b3c8..9765ed97a60a44 100644
--- a/Lib/test/test_unicode.py
+++ b/Lib/test/test_unicode.py
@@ -2370,15 +2370,19 @@ def test_expandtabs_optimization(self):
         self.assertIs(s.expandtabs(), s)
 
     def test_raiseMemError(self):
-        null_byte = 1
-        ascii_struct_size = sys.getsizeof("a") - len("a") - null_byte
-        compact_struct_size = sys.getsizeof("\xff") - len("\xff") - null_byte
+        asciifields = "nnb"
+        compactfields = asciifields + "nP"
+        ascii_struct_size = support.calcobjsize(asciifields)
+        compact_struct_size = support.calcobjsize(compactfields)
 
         for char in ('a', '\xe9', '\u20ac', '\U0010ffff'):
             code = ord(char)
-            if code < 0x100:
+            if code < 0x80:
                 char_size = 1  # sizeof(Py_UCS1)
                 struct_size = ascii_struct_size
+            elif code < 0x100:
+                char_size = 1  # sizeof(Py_UCS1)
+                struct_size = compact_struct_size
             elif code < 0x10000:
                 char_size = 2  # sizeof(Py_UCS2)
                 struct_size = compact_struct_size
@@ -2390,7 +2394,16 @@ def test_raiseMemError(self):
             # be allocatable, given enough memory.
             maxlen = ((sys.maxsize - struct_size) // char_size)
             alloc = lambda: char * maxlen
-            with self.subTest(char=char):
+            with self.subTest(
+                char=char,
+                struct_size=struct_size,
+                char_size=char_size
+            ):
+                # self-check
+                self.assertEqual(
+                    sys.getsizeof(char * 42),
+                    struct_size + (char_size * (42 + 1))
+                )
                 self.assertRaises(MemoryError, alloc)
                 self.assertRaises(MemoryError, alloc)
 
diff --git a/Misc/NEWS.d/next/Tests/2022-06-08-14-17-59.gh-issue-93575.Xb2LNB.rst b/Misc/NEWS.d/next/Tests/2022-06-08-14-17-59.gh-issue-93575.Xb2LNB.rst
new file mode 100644
index 00000000000000..98d15328a087ae
--- /dev/null
+++ b/Misc/NEWS.d/next/Tests/2022-06-08-14-17-59.gh-issue-93575.Xb2LNB.rst
@@ -0,0 +1,4 @@
+Fix issue with test_unicode test_raiseMemError. The test case now use
+``test.support.calcobjsize`` to calculate size of PyUnicode structs.
+:func:`sys.getsizeof` may return different size when string has UTF-8
+memory.