Dynamic switching base station (BS) to improve the utilization is very important for reducing the energy consumption. Most of existing dynamic switching BS schemes require the switching threshold to be set in advance. The fixed switching threshold, however, fails to apply in practical two-tier femtocell networks, for it is difficult to precisely model that it may result in different switching operations and subsequent worse system performance. In this paper, we propose an adaptive BS switching scheme to regulate the threshold considering the time varying characteristic of the traffic profile in the two-tier femtocell networks. We apply feedback control methodology to formulate the adaptive BS switching as an adaptive real-time control problem. Performance evaluation reveals that the solution has excellent stability behavior and physical reliability. And simulation results also indicate that the adaptive switching scheme can achieve better performance than the fixed one.