The mechanical properties of cells play important roles in regulating the physiological activities of cells and reflect the state of macro-organisms. Although many approaches are available for investigating the mechanical properties of cells, the fluidity of cytoplasm across cell boundaries makes characterizing the dynamics of mechanical properties of single cells exceedingly difficult. In this study, we present a single cell characterization method by modelling the dynamics of cellular mechanical properties measured with an atomic force microscope (AFM). The mechanical dynamics of a single cell system was described by a linear model with a mechanical stimulus as virtual input and mechanical property parameters as outputs. The dynamic mechanical properties of a single cell were characterized by the system matrix of the single cell system. The method was used to classify different types of cells, and the experimental results show that the proposed method outperformed conventional methods by achieving an average classification accuracy of over 90%. The developed method can be used to classify different cancer types according to the mechanical properties of tumour cells, which is of great significance for clinically assisted pathological diagnosis.