Risk classification is a fundamental task of insurance company to distinguish the consumers into different pricing categories based on the appropriate risk factors for a specific insurance coverage. In Taiwan automobile insurance market, the risk factors include gender, age, car type, claim coefficient except marital status. However, marital status is very common to be used as one of the major pricing factors in western countries such as U.S., Canada, and Germany. Empirical evidence indicates that, for various automobile insurance policies, marital status indeed contains unique characteristic information of risk. In this project, the focuses are: (1) Will marital status affect the existence of asymmetric information in automobile insurance market? (2) How do loss frequency and loss payment relate to marital status? (3) What is the correlation between car damage insurance and liability insurance after taking into account the role of marital status on the decision of purchasing and filing claims? The methodologies of this project are proposed by Chiappori and Salanie (2000) and Finkelstein and Poterba (2006). The former is used to analyze the information content of marital status, while the latter is for exploring the value of unused observables. In the first year, the data of automobile physical damage insurance from 2002-2006 is analyzed. In the second year, the data of third party voluntary liability insurance is explored. We expect to identify and to confirm the role of marital status in automobile insurance risk classification system. The policy implication will be provided.