diff --git a/budoux/__init__.py b/budoux/__init__.py
index 2e0dd903..0bf1f115 100644
--- a/budoux/__init__.py
+++ b/budoux/__init__.py
@@ -19,3 +19,4 @@
 
 Parser = parser.Parser
 load_default_japanese_parser = parser.load_default_japanese_parser
+DEFAULT_THRES = parser.DEFAULT_THRES
diff --git a/budoux/main.py b/budoux/main.py
index fa9f251c..600a787d 100644
--- a/budoux/main.py
+++ b/budoux/main.py
@@ -98,6 +98,12 @@ def parse_args(test: ArgList = None) -> argparse.Namespace:
       action="version",
       version="%(prog)s {}".format(budoux.__version__),
   )
+  parser.add_argument(
+      "--thres",
+      type=int,
+      default=budoux.DEFAULT_THRES,
+      help="threshold value to separate chunks (default: {})".format(
+          budoux.DEFAULT_THRES))
   if test is not None:
     return parser.parse_args(test)
   else:
@@ -116,13 +122,14 @@ def _main(test: ArgList = None):
       inputs = sys.stdin.read()
     else:
       inputs = args.text
-    res = parser.translate_html_string(inputs)
+    res = parser.translate_html_string(inputs, thres=args.thres)
   else:
     if args.text is None:
       inputs = [v.rstrip() for v in sys.stdin.readlines()]
     else:
       inputs = [v.rstrip() for v in args.text.splitlines()]
-    res = ["\n".join(res) for res in map(parser.parse, inputs)]
+    outputs = [parser.parse(sentence, thres=args.thres) for sentence in inputs]
+    res = ["\n".join(res) for res in outputs]
     ors = "\n" + args.delim + "\n"
     res = ors.join(res)
 
diff --git a/budoux/parser.py b/budoux/parser.py
index 66d3800a..651b58b5 100644
--- a/budoux/parser.py
+++ b/budoux/parser.py
@@ -22,6 +22,7 @@
 
 MODEL_DIR = os.path.join(os.path.dirname(__file__), 'models')
 PARENT_CSS_STYLE = 'word-break: keep-all; overflow-wrap: break-word;'
+DEFAULT_THRES = 1000
 with open(os.path.join(os.path.dirname(__file__), 'skip_nodes.json')) as f:
   SKIP_NODES: typing.Set[str] = set(json.load(f))
 
@@ -109,7 +110,7 @@ def __init__(self, model: typing.Dict[str, int]):
     """
     self.model = model
 
-  def parse(self, sentence: str, thres: int = 1000):
+  def parse(self, sentence: str, thres: int = DEFAULT_THRES):
     """Parses the input sentence and returns a list of semantic chunks.
 
     Args:
@@ -146,7 +147,7 @@ def parse(self, sentence: str, thres: int = 1000):
       p3 = p
     return chunks
 
-  def translate_html_string(self, html: str, thres: int = 1000):
+  def translate_html_string(self, html: str, thres: int = DEFAULT_THRES):
     """Translates the given HTML string with markups for semantic line breaks.
 
     Args:
diff --git a/tests/test_main.py b/tests/test_main.py
index 88a8c652..bc1766d8 100644
--- a/tests/test_main.py
+++ b/tests/test_main.py
@@ -106,6 +106,20 @@ def test_cmdargs_multi_html(self):
 
     self.assertEqual(cm.exception.code, 2)
 
+  def test_cmdargs_thres(self):
+    cmdargs = ['--thres=0', '今日はとても天気です。']
+    output_granular = main._main(cmdargs)
+    cmdargs = ['--thres=10000000', '今日はとても天気です。']
+    output_whole = main._main(cmdargs)
+    self.assertGreater(
+        len(output_granular), len(output_whole),
+        'Chunks should be more granular when a smaller threshold value is given.'
+    )
+    self.assertEqual(
+        ''.join(output_granular.split('\n')), ''.join(output_whole.split('\n')),
+        'The output sentence should be the same regardless of the threshold value.'
+    )
+
 
 class TestStdin(unittest.TestCase):