This paper studies the distributed filtering problem for time-delay systems under an adaptive event-trigger mechanism. An adaptive event trigger mechanism relies on the latest data released and an approximate value estimated by the filter, and the latest data from the neighbor filters. The threshold of the mechanism changes adaptively to minimize the filtering error in order to ensure the high performance of filtering and reduces the network resources consumption as much as possible. Firstly, this paper presents the sufficient conditions that stabilize the exponential mean-square of the filtering error manage system. Secondly, a Lyapunov function is constructed to analyze required performance when the filtering error manage system reaches the ${l_2