Analyzing the polymerase chain reaction (PCR) process by mathematical modelling is of importance in terms of uncovering the dynamic mechanism of PCR and predicting DNA amplification performance. Construction of an ideal PCR model, however, requires sufficient and accurate kinetic parameters, which cannot be easily obtained directly from experimental data. In this work, a genetic algorithm-based approach was introduced for optimizing kinetic parameters of PCR such as the rate constant of polymerase catalyst reaction. The fitted model agrees well with experimental data and predicts the DNA amplification yields as a function of cycle number. The kinetic parameters are not dependent on initial concentrations or the fragment types of DNA amplification. The model also allows us to predict the threshold cycle in real-time PCR, which is helpful for estimating the initial amount of DNA template and determining the optimal PCR reaction conditions.