The position information is usually needed when wireless sensor networks are applied for surveillance within certain region. Thus the localization of nodes is critical for WSNs. However, the present localization schemes mostly focus on static sensor networks, and in mobile WSNs, these schemes will not work well. This paper proposes a range-free Monte Carlo localization algorithm TSBMCL, which is extended from the Monte Carlo Boxed (MCB) scheme[1]. In this scheme, the well localized nodes are applied to aid other nodes for localization based on the sequential Monte Carlo method. The scheme could further improve the localization accuracy of the WSNs. We evaluate the performance of the scheme with comparison to MCB using MATLAB. The simulation results demonstrate that the localization scheme outperforms the MCB scheme.