The radioactive decay of 40K to 40Ar is the basis of isotope age determination of micaceous clay minerals formed during diagenesis. The difference in K–Ar ages between fine and coarse grained illite particles has been interpreted using detrital-authigenic components system, its crystallization history or post-crystallization diffusion. Yet another mechanism should also be considered: natural 40Ar recoil. Whether this recoil mechanism can result in a significant enough loss of 40Ar to provide observable decrease of K–Ar age of the finest illite crystallites at diagenetic temperatures – is the primary objective of this study which is based on molecular dynamics (MD) computer simulations. All the simulations were performed for the same kinetic energy (initial velocity) of the 40Ar atom, but for varying recoil angles that cover the entire range of their possible values. The results show that 40Ar recoil can lead to various deformations of the illite structure, often accompanied by the displacement of OH groups or breaking of the Si–O bonds. Depending on the recoil angle, there are four possible final positions of the 40Ar atom with respect to the 2:1 layer at the end of the simulation: it can remain in the interlayer space or end up in the closest tetrahedral, octahedral or the opposite tetrahedral sheet. No simulation angles were found for which the 40Ar atom after recoil passes completely through the 2:1 layer. The energy barrier for 40Ar passing through the hexagonal cavity from the tetrahedral sheet into the interlayer was calculated to be 17 kcal/mol. This reaction is strongly exothermic, therefore there is almost no possibility for 40Ar to remain in the tetrahedral sheet of the 2:1 layer over geological time periods. It will either leave the crystal, if close enough to the edge, or return to the interlayer space. On the other hand, if 40Ar ends up in the octahedral sheet after recoil, a substantially higher energy barrier of 55 kcal/mol prevents it from leaving the TOT layer over geological time. Based on the results of MD simulations, the estimates of the potential effect of 40Ar recoil on the K–Ar dating of illite show that some of 40Ar is lost and the loss is substantially dependent on the crystallite dimensions. The 40Ar loss can vary from 10% for the finest crystallites (two 2:1 layers thickness and <0.02 μm in diameter) to close to zero for the thickest and largest (in the ab plane) ones. Because the decrease of the K–Ar estimated age is approximately proportional to the 40Ar loss, the finer crystallites show lower apparent age than the coarser ones, although the age of crystallization is assumed equal for all the crystallites. From the model it is also clear that the lack of K removal from illite fringes (potentially Ar-free) strongly increases the apparent age differences among crystallites of different size.