The horizontal stratification occurrence in WNCR plays an important role in many transients such as loss of coolant accidents and loss of residual heat removal by influencing the liquid mass repartition, the natural circulation and reflux condenser modes, and the break or leak flowrate. A benchmark activity on horizontal flow prediction was conducted in the framework of FONESYS. FONESYS is a network among code developers who share the common objective to strengthen current technology. The aim of the network is to highlight the capabilities and the robustness as well as the limitations of current SYS-TH codes to predict the main phenomena during transient scenarios in nuclear reactors for safety issues. One-hundred-twenty-six runs performed with the Two-Phase Test Facility (TPTF) were considered. Tests were selected aiming to support and complement an ongoing activity on the scalability of codes' closure laws, currently focusing on the horizontal stratification criteria and the onset of droplet entrainment and entrainment fraction criteria. Therefore, experiments performed with the 4-inch and 8-inch tests sections, at pressure ranging from 30 bar to 118 bar, in different flow regimes were considered. Focus was put on the capability of ATHLET, CATHARE 3, MARS-KS, RELAP5, RELAP5-3D, SPACE and TRACE in predicting void fractions, the transition to slug flow and the onset of droplet entrainment. Overall, the benchmark results show reasonably good capabilities of codes in predicting horizontal flow. However, some noticeable limitations emerged related to flow regime transition criteria, interfacial friction, and to the prediction of non-established flow. The scalability of closure laws and possible ways for improving them are discussed.