This paper proposed a method that can automatically detect syntax errors in Chinese sentences. The algorithm for identifying syntax errors proposed in this study is known as KNGED, which uses a large database of rules to identify whether syntax errors exist in a sentence. The rules were generated either manually or automatically. This paper further proposed an algorithm for identifying the type of error that a sentence contained. Experimental results shown that the false positive rate and F1-measure of the proposed method for detecting syntax errors in Chinese sentences are 0.90 and 0.65.