diff --git a/pkuseg/__init__.py b/pkuseg/__init__.py
index becad8c..011c5dd 100644
--- a/pkuseg/__init__.py
+++ b/pkuseg/__init__.py
@@ -57,16 +57,23 @@ def solve(self, txt):
             now = self.trie
             j = i
             found = False
+            last_word_idx = -1 # 表示从当前位置i往后匹配，最长匹配词词尾的idx
             while True:
                 c = txt[j]
-                if not c in now.children:
+                if not c in now.children and last_word_idx != -1:
+                    found = True
+                    break
+                if not c in now.children and last_word_idx == -1:
                     break
                 now = now.children[c]
-                j += 1
                 if now.isword:
-                    found = True
+                    last_word_idx = j
+                j += 1
+                if j == l and last_word_idx == -1:
                     break
-                if j == l:
+                if j == l and last_word_idx != -1 :
+                    j = last_word_idx + 1
+                    found = True
                     break
             if found:
                 if last != i: