Abstarct: By increasing electronic text resources on the World Wide Web, the need to accuratly, fast and easy access to more information is vital. In order to find fast and suitable the user's documents, complete reading great literature is inappropriate. The automatic summarization system play an important role and is also a solution to the problem of information overload. Summarization is actually a compact and accurate text input representation so that the output text represent important concepts of the input text. Some of the applications of summarization are automatic news summarization while sending them via e-mail or SMS. other applications can be named research-oriented summarization, business-oriented summarization, information retrival system,industry communication ,editing and filtering systems and web page summarization which are to be displayed on mobile screens. There are two major challenges in designing a system for Persian language summarization. The first challenge is continuity between some symptoms and various forms of words and the second is selecting the most important continuous sentence that are to be present in the summary. In this thesis, duo to the challenges in the summarization of Persian language , Preprocessing step is completely done. Extacting related and continuous sentence in proposed summarization method, requires related and accurate keywords and therefore we have proposed a method for extracting co-occurrences of words and event detection of text for persian news summarization. In Proposed method, news tixtle, news text and compression rate are received from users and produce summary according to compression rate and related news tixtle. The Proposed method is an unsupervised and statistical method and is a kind of the extraction summary. In order to evaluate proposed method, measures of precision, recall and comparison to human summary and to comparison to FarsiSum are used. Evaluations have be done on set of persion news of Hamshahry databaxse in different scope and domain. According to our experiment ,our system has the best performance among other Persian summarization system(%81/66),the closest to human summaries and it also has a better performance to comared FursiSum system.