diff --git a/ChangeLog.md b/ChangeLog.md
index 9ac88bebe9f40..9ef5b78559ff0 100644
--- a/ChangeLog.md
+++ b/ChangeLog.md
@@ -67,9 +67,6 @@ See docs/process.md for more on how version tagging works.
 - Added SAFE_HEAP=2 option which tests safe heap behavior for wasm-only builds
   (allowing unaligned memory accesses, which would not work in Wasm2JS but in
    wasm would be correct but potentially slow).
-- Added support for specifying the text encoding to be used in response filenames
-  by passing the encoding as a file suffix (e.g. "a.rsp.utf-8" or "a.rsp.cp1252").
-  If not specified, the encoding is autodetected. (#15406, #15292)
 
 2.0.31 - 10/01/2021
 -------------------
diff --git a/tests/test_other.py b/tests/test_other.py
index 24bfd0f459856..ab352aca6fc10 100644
--- a/tests/test_other.py
+++ b/tests/test_other.py
@@ -10737,21 +10737,6 @@ def create_o(name, i):
     self.run_process(building.get_command_with_possible_response_file([EMCC, 'main.c'] + files))
     self.assertContained(str(count * (count - 1) // 2), self.run_js('a.out.js'))
 
-  # Tests that the filename suffix of the response files can be used to detect which encoding the file is.
-  def test_response_file_encoding(self):
-    open('äö.c', 'w').write('int main(){}')
-
-    open('a.rsp', 'w', encoding='utf-8').write('äö.c') # Write a response file with unicode contents ...
-    self.run_process([EMCC, '@a.rsp']) # ... and test that in the absence of a file suffix, it is autodetected to utf-8.
-
-    open('a.rsp.cp437', 'w', encoding='cp437').write('äö.c') # Write a response file with Windows CP-437 encoding ...
-    self.run_process([EMCC, '@a.rsp.cp437']) # ... and test that with the explicit suffix present, it is properly decoded
-
-    if WINDOWS:
-      # This test assumes that the default system encoding is CP-1252.
-      open('a.rsp', 'w', encoding='cp1252').write('äö.c') # Write a response file with Windows CP-1252 encoding ...
-      self.run_process([EMCC, '@a.rsp']) # ... and test that it is properly autodetected.
-
   def test_output_name_collision(self):
     # Ensure that the seconday filenames never collide with the primary output filename
     # In this case we explcitly ask for JS to be ceated in a file with the `.wasm` suffix.
diff --git a/tools/response_file.py b/tools/response_file.py
index 1e25997b606a6..a6013f76c51cb 100644
--- a/tools/response_file.py
+++ b/tools/response_file.py
@@ -13,16 +13,13 @@
 DEBUG = int(os.environ.get('EMCC_DEBUG', '0'))
 
 
-def create_response_file(args, directory, suffix='.rsp.utf-8'):
+def create_response_file(args, directory):
   """Routes the given cmdline param list in args into a new response file and
   returns the filename to it.
 
-  By default the returned filename has a suffix '.rsp.utf-8'. Pass a suffix parameter to override.
+  The returned filename has a suffix '.rsp'.
   """
-
-  assert suffix.startswith('.')
-
-  response_fd, response_filename = tempfile.mkstemp(prefix='emscripten_', suffix=suffix, dir=directory, text=True)
+  response_fd, response_filename = tempfile.mkstemp(prefix='emscripten_', suffix='.rsp', dir=directory, text=True)
 
   # Backslashes and other special chars need to be escaped in the response file.
   escape_chars = ['\\', '\"']
@@ -44,12 +41,16 @@ def escape(arg):
       arg = '"%s"' % arg
     contents += arg + '\n'
 
-  # Decide the encoding of the generated file based on the requested file suffix
-  if suffix.count('.') == 2:
-    # Use the encoding specified in the suffix of the response file
-    encoding = suffix.split('.')[2]
-  else:
+  # When writing windows repsonse files force the encoding to UTF8 which we know
+  # that llvm tools understand.  Without this, we get whatever the default codepage
+  # might be.
+  # See: https://github.com/llvm/llvm-project/blob/3f3d1c901d7abcc5b91468335679b1b27d8a02dd/llvm/include/llvm/Support/Program.h#L168-L170
+  # And: https://github.com/llvm/llvm-project/blob/63d16d06f5b8f71382033b5ea4aa668f8150817a/clang/include/clang/Driver/Job.h#L58-L69
+  # TODO(sbc): Should we also force utf-8 on non-windows?
+  if WINDOWS:
     encoding = 'utf-8'
+  else:
+    encoding = None
 
   with os.fdopen(response_fd, 'w', encoding=encoding) as f:
     f.write(contents)
@@ -76,25 +77,8 @@ def read_response_file(response_filename):
   if not os.path.exists(response_filename):
     raise IOError("response file not found: %s" % response_filename)
 
-  # Guess encoding based on the file suffix
-  components = os.path.basename(response_filename).split('.')
-  encoding_suffix = components[-1].lower()
-  if len(components) > 1 and (encoding_suffix.startswith('utf') or encoding_suffix.startswith('cp') or encoding_suffix.startswith('iso') or encoding_suffix in ['ascii', 'latin-1']):
-    guessed_encoding = encoding_suffix
-  else:
-    guessed_encoding = 'utf-8'
-
-  try:
-    # First try with the guessed encoding
-    with open(response_filename, encoding=guessed_encoding) as f:
-      args = f.read()
-  except (ValueError, LookupError): # UnicodeDecodeError is a subclass of ValueError, and Python raises either a ValueError or a UnicodeDecodeError on decode errors. LookupError is raised if guessed encoding is not an encoding.
-    if DEBUG:
-      logging.warning('Failed to parse response file ' + response_filename + ' with guessed encoding "' + guessed_encoding + '". Trying default system encoding...')
-    # If that fails, try with the default system encoding
-    with open(response_filename) as f:
-      args = f.read()
-
+  with open(response_filename) as f:
+    args = f.read()
   args = shlex.split(args)
 
   if DEBUG: