ABSTRACTMolecular Dynamics (MD) is a very powerful tool for studying displacement cascades initiated by the neutrons when they interact with matter and thus evaluate the primary damage. The mean number of point defects created can be obtained with a fair standard error with a reasonable number of cascade simulations (10 to 20 [1]), however other cascades characteristics (spatial distribution, size and amount of defect clusters …) display a huge variability. Therefore, they may need to be studied using faster methods such as the Binary Collision Approximation (BCA) which is several order of magnitude less time consuming. We have investigated the point defect distributions subsequent to atomic collision cascades by both MD (using EAM potentials for Fe) and its BCA. MD and its BCA lead to comparable point defect predictions. The significant similarities and differences are discussed.