Time-domain synthesis is used when it is necessary to determine the system transfer function whose time-domain response approximates a given response. Typically, it is used in the design of systems approximating a given impulse response or step response shape, as well as in the design of systems ensuring low time-domain distortion. It is less known that the time-domain synthesis can be used in the design of selective filters. In this study, the design of selective filters with linear phase whose design is based on the time-domain synthesis is considered. In the first place, an existing method, which is based on the approximation of empirically obtained impulse response, is improved. Furthermore, new methods are developed for the design of filters whose impulse responses approximate the sinc pulse and the Dirac function. By using these methods, new classes of filters are obtained. These classes are suitable for practical applications. Additionally, they provide a new insight into the behavior of the systems. The design based on the approximation of the sinc pulse or the Dirac function is straight forward in the sense the designer does not need to pay attention on filters magnitude or phase response. However, for the case where it is required, the procedure is developed for design of filters approximating a given magnitude and the linear phase response.