A new mathematical model for generating ECG signal is presented in this paper. The generation is based on geometrical properties of real ECG signals instead of various artificial function used in commonly applicated models. Variation of parameters allows arbitrary setting of particular waves of PQRST complex. This way arbitrary PQRST signals typical for various heart diseases can be modelled and generated with high fidelity. Moreover, parameter variation allows generate different waveform for each subsequent heartbeat without mixing PQRST waves order. The elementary trigonometric function and Gaussian monopulse have been used for modelling each particular wave. Adding a noise and/or respiratory signal the final synthetized artificial signal can be apply for various signal processing method testing. The model was tested by comparison of synthetized patterns against patterns generated by LabVIEW Biomedical Toolkit, while the parameters of the model are found using differential evolution algorithm. Professor. The full paper will contain some new improvements of the proposed model and its comparison with real ECG from patients’ database.