The quantitative analysis of live cells is a key issue in evaluating biological processes. The current clinical practice involves the application of a tedious and time consuming manual tracking procedure on large amount of data. As a result, automatic tracking systems are currently developed and evaluated. However, problems caused by cellular division, agglomeration, Brownian motion and topology changes are difficult issues that have to be accommodated by automatic tracking techniques. In this paper, we detail the development of a fully automated multi-target tracking system that is able to deal with Brownian motion and cellular division. During the tracking process our approach includes the neighbourhood relationship and motion history to enforce the cellular tracking continuity in the spatial and temporal domain. The experimental results reported in this paper indicate that our method is able to accurately track cellular structures in time-lapse data.