Recent publications have called into question the accuracy of reference tenth‐value layer (TVL) data cited in official reports for linac primary concrete barriers. Doubts have arisen based on both experimental and theoretical evidence. Most of the standard reference TVL values trace back to a publication that appeared in 1984 that used beam spectra that are not representative of modern linacs. This study reports a new set of TVL data for concrete based on modern linac beam spectra and a definition of the barrier transmission that is consistent with its use in shielding calculations. TVL values have been computed for concrete using Monte Carlo simulation for beam energies of 4, 6, 10, 15, and 18 MV. The barrier transmission depends on the field size at the barrier and the distance from the distal surface of the barrier to the point of observation. The TVL values reported here lead to barrier transmission values that are up to a factor of 4 larger than those in official reports. The air kerma rate beyond the barrier does not obey an inverse square law as the barrier now acts like a new (non‐point) source of radiation. For distance greater than 0.3 m from the distal side of the barrier, inverse square predictions of the air kerma rate are low by up to a factor of 2. The average energy of the transmitted photons declines rapidly for all beam energies with increasing barrier thickness up to a thickness of about 50 cm and then slowly increases with increasing thickness.