AbstractA benchmark experiment on void effects in a bulk shield assembly made of type 316 stainless steel was carried out, and prediction capabilities of these void effects using three transport calculation codes, i.e. DOT, MCNP and MVP was investigated. It is found that continuous energy Monte Carlo calculations with MCNP and MVP predict almost perfectly the void effects. Multi-group SN calculations by DOT may contain uncertainties in the prediction of the void effects when the self-shielding correction from the cross section is disregarded. The use of less angular quadratures such as S8 is also a cause of uncertainties due to the ray-effect radiated from the secondary source point at the beginning of the void space appearing.