The SQLite is applied by a number of Instant Messaging (IM) softwares for history data storing. In the process of IM forensics, to impede the investigation of the judiciary, the important SQLite data are often hidden, deleted or overwritten by criminals. The current data recovery methods are inefficient and cannot extract the overwritten data. A content carving algorithm based on SQLite was proposed to resolve the above problems. According to SQLite database storage characteristics and data deletion mechanism, the free domain was regarded as units to form idle list, the page content was used to grain structural units of engraving, and the scattered blocks of data were spliced efficiently on the basis of the position of data overwritten. The experimental results show that the propsed SQLite content carving algorithm can effectively recover IM data in local and mobile terminals; the recovery rate reaches 100% when the database is not damaged, and the recovery rate still reaches about 50% when the delete area is overwritten in different degrees.