Health risks from exposure to high doses of ionizing radiation are well characterized from epidemiological studies. Uncertainty and controversy remain for extension of these risks to the low doses and low dose rates of particular relevance in the workplace, in medical diagnostics and screening, and from background radiations. In order to make such extrapolations, a number of concepts have been developed for radiation protection, partly on the basis of assumed processes in the mechanisms of radiation carcinogenesis. Included amongst these are the assumptions of a linear no-threshold dose response and simple scaling factors for dose rate and radiation quality. With a progressive reduction in recommended dose limits over the past half century, these approaches have had considerable success in protecting humans. But do they go far enough or, conversely, are they overprotective? Four selected underlying aspects are considered. It is concluded that (1) even the lowest dose of radiation has the capability to cause complex DNA damage that can lead to a variety of permanent cellular changes; (2) the unique clustered characteristics of radiation damage, even at very low doses, enable it to stand out above the much larger quantity of endogenous DNA damage; (3) although a chromosome aberration may represent the rate-limiting initiating event for carcinogenesis, as is often assumed, direct evidence is still lacking; and (4) the extensive influence that dicentric aberrations have had on guiding extrapolations for radiation protection may be substantially misleading. Finally, some comments are offered on aspects that lie outside the current paradigm.