Presently, many ideas have been proposed for constructing zonal models of the European transmission network. Most of them are based on transmission network reduction and bus clustering, which requires the ex-ante knowledge on the details (impedance, interconnections, etc.) of the network’s elements. Generally, the information is not publicly disclosed. This paper describes a method for serving the purpose, which only requires the ex-post (i.e., published) inter-zonal power flow data as input. It performs inverse modeling on the ex-post data to reveal the virtual reactance of the interconnectors in the zonal model. The reactance values are then used for computing the zonal Power Transfer Distribution Factors (PTDFs) of the model, which is the best representation of the loop flows. A large number of such zonal PTDFs could exist in one zonal model because of its dependency on the Generation Shift Keys (GSKs). Theoretically, the GSKs could exist as an infinite set of unique combinations but only a finite number of them applied in real system operations. As a solution, the authors propose a heuristic approach for grouping the ex-post power flow cases into clusters, and find the zonal PTDF for each of them; in order to effectively reduce the domain of the zonal PTDFs and enhance the tractability of the solution. The identified zonal PTDFs meaningfully represent the zonal model. Additionally, they can also be used for studying the best and worst case scenarios of cross-border security-constrained economic dispatch.