The paper provides a detailed analysis of the coefficient quantization error on the amplitude response for a general second-order section. Seven classes of two-pair structures and the canonic direct form structure are used to realize the sect.ion. The amplitude response sensitivity of the second-order sections obtained for sixth-order low-pass digital filters with variable bandwidth for the four types of classic approximation is computed. Finally, in order to minimize the deterioration due to coefficient quantization in tho n.mplitude-frequency response of a digital filter, it is proposed that tho realization of the digital filter should be by cascade connection of second-order sections chosen by using the above results. This deterioration is then evaluated by computer simulation.