Abstract. The tracking of individual cells in time-lapse microscopy facilitates the assessment of certain characteristics of different cell types. Since manual tracking of an adequate number of cells over a considerable number of frames is tedious and sometimes not feasible, there is a vital interest in automated methods. We present a rather minimalistic approach for the tracking of unstained cells in cell culture assays. The proposed approach comprises background subtraction, an object detection method based on discrete geometrical feature analysis together with a validation of the resulting graph-structures. The main advantage of this approach lies in its computational efficiency.