Beam loss criterion such as 1 W/m for an uniformly distributed loss has been applied at high power proton or heavy ion accelerator. It has been accepted based on the exposure level due to radioactive accelerator components during a hands-on maintenance and is a very important factor in shielding analysis of such high power, high energy accelerators. Well-developed Monte Carlo codes and inventory codes, which have been used for an activity calculation, were confirmed by benchmarking of published experimental data. The modular style method (PHITS+DChain-SP) and all-in-one style method (Fluka) using Monte Carlo code were applied to verify the criterion. The beam loss at bulk target and one-point loss at beam pipe and a uniformly distributed loss were simulated. The dose distribution and the decay scheme were compared with the control level of a hands-on maintenance. It was proved that more considerations were required instead of taking 1 W/m simply as the beam loss criterion. Especially the energy dependency effect was discussed mainly.