Abstract: How to improve the efficiency of data access is always a hot topic in the research area of mobile opportunistic networks. Traditional cooperative caching techniques are commonly used to improve the performance of data access. However, the strongly independent mobility and limited contact duration of the mobile nodes render these traditional caching schemes inefficient. Firstly, a new metric, called as node important degree, is proposed to determine which node is more important to cooperative data caching. Based on this metric, a greedy algorithm is used to select initial cache nodes, and subsequently the cache data will be redistributed among these cache nodes actively as they meet each other. A novel data fragmenting strategy is suggested to adapt to the limited contact duration between nodes with the aim to make our protocol suitable for short-duration contact between cache nodes. In order to solve the coupon collector’s problem in data recovery, a randomly linear network coding method is used to encode the data fragmentations. Moreover, we describe an adaptive caching bound calculation method for each mobile node to limit the amount of data it caches, which is helpful to the rational utilization of cache space. Experimental results show that our suggested cooperative caching protocol can significantly improve the efficiency of data access in mobile opportunistic networks.