A new and efficient technique of decomposition impedance spectra is presented. It is particularly suitable for analyzing data obtained from ceramic samples, when a simple model, e.g. with four free parameters, is not adequate. An initial impedance model is obtained by a parametric fitting, such as ISGP [Hershkovitz, Baltianski and Tsur, Solid State Ionics, 188, (2011) 104]. It results in an analytic distribution function of relaxation times combined of known functions like Gaussians. Peaks in the distribution function represent processes. For each process, asymptotic parameters of effective resistance and capacitance are determined first, to allow extracting a residual admittance. The residual admittance is further processed as will be described in the presentation. This is demonstrated using IS data from BaTiO3 ceramics over the range 175 to 325 ºC and 10 mHz to 30 MHz. The parameters are obtained for both the grain boundaries and bulk.