The US Environmental Protection Agency’s Radiation Protection Division (EPA/RPD) appreciates the opportunity to comment on the 2006 Draft Recommendations of the ICRP and commends the Commission for its continued receptivity to broad stakeholder input in the development of the next generation of ICRP guidance. The current draft represents considerable progress and improvement since the 2005 draft. EPA participated in the development of the comments recently posted by the Federal Guidance Subcommittee of the US Interagency Steering Committee on Radiation Standards (ISCORS) and we would like to endorse those comments by reference. Additional comments of particular interest to EPA are listed below. Constraints EPA disagrees with the frequently stated assertion that dose constraints are the fundamental level of protection in the system of radiation protection. From a regulatory perspective, the fundamental level of protection must be the dose (or risk) limit. We agree that the use of constraints is an effective tool for apportioning the total dose limit across multiple sources so as to ensure that the sum of the contributions from individual sources is below the relevant dose limit. Once demystified, the constraint appears to be nothing more than a common sense step in the overall optimization process. Thus care should be taken that the role of the constraint is not elevated out of proportion to its practical utility as an intermediate step in the process of controlling sources of radiation exposure. Further, we note that there is still confusion over the use of constraints in existing and emergency situations. For example, in emergency situations, the implication that averted doses (used as protective action guides) are constraints is, we believe, incorrect. We therefore urge the Commission to consider consolidating and clarifying all the information on constraints into one section of the Recommendations. Collective Dose In Paragraph 57, the ICRP recommends that it is not appropriate to project health effects due to collective doses consisting of very low individual doses to large numbers of people. Two arguments might be put forward for this position. First, the validity of the LNT model at low doses is so questionable as to render the result misleading or meaningless. Second, the individual doses are so low as to make the individual risks trivial; and the sum of trivial risks is necessarily trivial. Neither of these arguments is compelling in our view. As elucidated in ICRP Publication 99, the LNT model, while certainly not proved, has a great deal of scientific support and remains highly plausible. The idea that the sum of trivial risks is itself trivial has no logical basis. For example, the individual risks to people irradiated outside the Chernobyl exclusion zone may indeed be trivial, but this does not mean that the expected number of cancers induced is necessarily trivial. We agree that the concept of collective dose can be misused, e.g., in estimating health effects to future populations whose characteristics are highly speculative. However, there is sufficient scientific support for LNT that it is reasonable to project health effects in current populations, even at very low doses, so long as the uncertainty regarding LNT is noted and so long as the magnitude of the small individual risks are properly put into perspective by comparing them with risks from background radiation and/or overall cancer risks in the population. Therefore, the ICRP statement on misuse of collective dose should be modified to say that cancers projected from low radiation doses to large populations should not be presented without both noting the uncertainties in the assumptions underlying the calculations and putting the estimated risks and health impacts into proper perspective. Tissue Weighting Factors The draft recommendations continue to employ a tissue weighting factor of 0.01 for bone. This seriously overstates the importance of bone as a target tissue. As pointed out in a paper by Puskin et al. (Health Phys. 63: 579-580; 1992), the overestimate arises from confusion over skeletal dosimetry. More specifically, the ICRP risk estimate reflects the average dose to the skeleton from deposited Ra-224 rather than the dose to the endosteum, which is the presumed target for cancer induction. We recognize that the ICRP may not want to assign tissue weighting factors of less than 1 percent. However, this issue should be addressed in material relating to the risk estimates underlying the tissue weighting factors and in the treatment of detriment.