<p>The detection of plastic anti-personnel landmines poses a significant challenge. The presence of explosive devices near occupied nuclear facility does not meet the IAEA safety standards and nuclear security guidance. Widely used detection techniques rely on identifying metal components, making them ineffective for non-metallic explosives. In this work, Monte Carlo simulations were performed using the MCNP6 code to design and evaluate a neutron-based detection system capable of identifying such landmines. A Californium-252, point neutron source, was employed due to its spontaneous fission properties and high neutron yield. The model is based on the thermal neutron backscattering technique. This method depends on the difference in hydrogen concentration between landmines and the surrounding environment, as plastic explosives typically contain higher amounts of hydrogen. The signal-to-background (S/B) ratio of backscattered neutrons was calculated to assess the system’s ability to differentiate between areas with and without landmines. Sensitivity tests were carried out under different conditions to evaluate their performance in realistic scenarios. These tests aimed to analyse the model response to variations in burial depths, soil compositions, and landmine types, providing a better understanding of their detection capabilities and limitations.</p>