diff --git a/Lib/test/test_bigmem.py b/Lib/test/test_bigmem.py
index 7f53379a952dfc..3c09cdb8640890 100644
--- a/Lib/test/test_bigmem.py
+++ b/Lib/test/test_bigmem.py
@@ -901,7 +901,7 @@ def test_repeat_small(self, size):
     def test_repeat_large(self, size):
         return self.basic_test_repeat(size)
 
-    @bigmemtest(size=_1G - 1, memuse=12)
+    @bigmemtest(size=_1G - 1, memuse=pointer_size * 3)
     def test_repeat_large_2(self, size):
         return self.basic_test_repeat(size)
 
diff --git a/Lib/test/test_hashlib.py b/Lib/test/test_hashlib.py
index c7d7b801ac732a..fa13b553c88f86 100644
--- a/Lib/test/test_hashlib.py
+++ b/Lib/test/test_hashlib.py
@@ -626,12 +626,12 @@ def test_case_md5_2(self):
         )
 
     @unittest.skipIf(sys.maxsize < _4G + 5, 'test cannot run on 32-bit systems')
-    @bigmemtest(size=_4G + 5, memuse=1, dry_run=False)
+    @bigmemtest(size=_4G + 5, memuse=2, dry_run=False)
     def test_case_md5_huge(self, size):
         self.check('md5', b'A'*size, 'c9af2dff37468ce5dfee8f2cfc0a9c6d')
 
     @unittest.skipIf(sys.maxsize < _4G - 1, 'test cannot run on 32-bit systems')
-    @bigmemtest(size=_4G - 1, memuse=1, dry_run=False)
+    @bigmemtest(size=_4G - 1, memuse=2, dry_run=False)
     def test_case_md5_uintmax(self, size):
         self.check('md5', b'A'*size, '28138d306ff1b8281f1a9067e1a1a2b3')
 
diff --git a/Lib/test/test_re.py b/Lib/test/test_re.py
index f6e797b3785dbe..fc052a2f9329ec 100644
--- a/Lib/test/test_re.py
+++ b/Lib/test/test_re.py
@@ -2026,7 +2026,7 @@ def test_large_search(self, size):
 
     # The huge memuse is because of re.sub() using a list and a join()
     # to create the replacement result.
-    @bigmemtest(size=_2G, memuse=16 + 2)
+    @bigmemtest(size=_2G, memuse=16 + 3)
     def test_large_subn(self, size):
         # Issue #10182: indices were 32-bit-truncated.
         s = 'a' * size
diff --git a/Lib/test/test_tcl.py b/Lib/test/test_tcl.py
index 70731d3222ced9..f6e49840d063bc 100644
--- a/Lib/test/test_tcl.py
+++ b/Lib/test/test_tcl.py
@@ -789,7 +789,7 @@ def test_huge_string_call(self, size):
 
     @support.cpython_only
     @unittest.skipUnless(INT_MAX < PY_SSIZE_T_MAX, "needs UINT_MAX < SIZE_MAX")
-    @support.bigmemtest(size=INT_MAX + 1, memuse=2, dry_run=False)
+    @support.bigmemtest(size=INT_MAX + 1, memuse=3, dry_run=False)
     def test_huge_string_builtins(self, size):
         tk = self.interp.tk
         value = '1' + ' ' * size
diff --git a/Lib/test/test_zlib.py b/Lib/test/test_zlib.py
index 4d97fe56f3a094..32add7bc1e92f7 100644
--- a/Lib/test/test_zlib.py
+++ b/Lib/test/test_zlib.py
@@ -982,7 +982,7 @@ def testEOFError(self):
         self.assertRaises(EOFError, zlibd.decompress, b"")
 
     @support.skip_if_pgo_task
-    @bigmemtest(size=_4G + 100, memuse=3.3)
+    @bigmemtest(size=_4G + 100, memuse=4.5)
     def testDecompress4G(self, size):
         # "Test zlib._ZlibDecompressor.decompress() with >4GiB input"
         blocksize = min(10 * 1024 * 1024, size)