diff --git a/rasa/nlu/tokenizers/jieba_tokenizer.py b/rasa/nlu/tokenizers/jieba_tokenizer.py
index e0631c19f3cb..d6fc90ac5665 100644
--- a/rasa/nlu/tokenizers/jieba_tokenizer.py
+++ b/rasa/nlu/tokenizers/jieba_tokenizer.py
@@ -101,7 +101,15 @@ def tokenize(self, message: Message, attribute: Text) -> List[Token]:
         text = message.get(attribute)
 
         tokenized = jieba.tokenize(text)
-        tokens = [Token(word, start) for (word, start, end) in tokenized]
+        tokens = []
+        current_position = 0
+        for word, start, end in tokenized:
+            if word.strip() == "":
+                continue
+            word_start = text.find(word, current_position)
+            word_end = word_start + len(word)
+            tokens.append(Token(word, word_start, word_end))
+            current_position = word_end
 
         return self._apply_token_pattern(tokens)
 
diff --git a/tests/nlu/tokenizers/test_jieba_tokenizer.py b/tests/nlu/tokenizers/test_jieba_tokenizer.py
index c0628f901a87..f2d93471d84a 100644
--- a/tests/nlu/tokenizers/test_jieba_tokenizer.py
+++ b/tests/nlu/tokenizers/test_jieba_tokenizer.py
@@ -37,6 +37,11 @@ def create_jieba(config: Optional[Dict] = None) -> JiebaTokenizer:
             ["Micheal", "你好", "吗", "？"],
             [(0, 7), (7, 9), (9, 10), (10, 11)],
         ),
+        (
+            "安装 rasa 应用",
+            ["安装", "rasa", "应用"],
+            [(0, 2), (3, 7), (8, 10)],
+        ),
     ],
 )
 def test_jieba(text, expected_tokens, expected_indices):