The thermal neutron activation measurements carried out over many years in Hiroshima and Nagasaki have been the subject of ongoing debate in recent years because they indicate that current DS86 neutron doses may have been significantly underestimated in Hiroshima. Long-lived neutron activation products, 60Co, 152Eu, 154Eu and 36Cl, which are still detectable today using modern analytical techniques, appear to indicate that DS86 calculated thermal neutron activation products decrease with distance more rapidly than the measured values. The latest thermal neutron activation measurements have been collated and a new relationship for the measured to calculated (M/C) ratio of induced activity has been derived as a function of slant range. This indicates a stronger dependence of M/C on slant range than previously derived by Straume et al (1992 Health Phys. 63 421-6) and emphasises even more the discrepancy between measured and calculated (DS86) neutron doses at distances beyond 1 km. While the main body of thermal neutron activation data appears to support a significant increase in the DS86 neutron dose component in Hiroshima, there are some thermal neutron activation measurements and some very recent fast neutron activation measurements which suggest that the discrepancy may not be so great. The extent of the required revision to the neutron component of the DS86 dosimetry remains the subject of ongoing new neutron activation measurements and re-analysis of existing published measurements. A companion paper considers the impact on radiation risk estimates of possible modifications to the DS86 dosimetry system on the basis of a broad range of interpretations of the neutron activation data.