There is a systematic discrepancy between the measured radionuclide activation yields
in materials exposed to neutrons from the Hiroshima atomic bomb (A-bomb) and the calculated activation based on the Dosimetry System 1986 (DS86). The data at lkm are 2-10-times greater than the calculation and this implies that the actual level of exposure from the A-bomb must be higher than previously thought. In this paper, these activity data, including 152Eu, 60Co, 36C1 and 32P, were analyzed using a computer code (the Monte Carlo neutron and photon (MCNP) transport code system) to investigate the cause of the discrepancy. We present a possible mechanism of the neutron emission in Hiroshima which can reproduce the activity yields of 152Eu, 60Co, 36C1 and 32P assuming a small amount of leakage in the lateral direction (leakage except that in the direction to the hypocenter area). The DS86 calculation did not take such leakage into account. In addition to this leakage, we assume increase of the burst height about 90 m to adjust the activity yields at relatively short ground distances. Taking these two assumptions into account and using a simple bomb model and leakage, it is shown that our calculation reproduces reasonably both the slopes and the absolute yield data of 152Eu, 32P, 60Co and 36C1 within about 1 km ground range. Over the ground range of about 1 km, it seems that further consideration is necessary concerning not only the calculation but also the data themselves to explain the DS86 discrepancy.