The human evidence on radiation damage to the individual developing in utero is confined to mental impairment and carcinogenesis. New evidence is becoming available about levels of mental impairment of direct interest to radiological protection, but as yet no framework of understanding exists to allow quantitative predictions for the purposes of radiological protection. There is general agreement that malignant disease has been increased following antenatal radiography but no unanimity yet in concluding that irradiation was the main causal factor: reasons are given for accepting that radiography was the cause. Recent increases in biological understanding suggest why maldevelopment is not to be expected after irradiation of the conceptus. A clonal hypothesis for organogenesis provides a reasonable explanation for quantitative aspects of experimental observations on teratogenesis by ionising radiation, including the commonly found highly curvilinear dose-response relationship, the occurrence of so-called critical stages of sensitivity after exposures of a few hundred roentgens, and the reduction in frequency of induced abnormality with protraction of exposure. Clonal hypotheses predict that there will be a virtual threshold for polycystic (non-stochastic) forms of radiation damage. It may be misguided to adopt a linear dose-response relationship for deriving risk estimates for the practical purposes of radiological protection unless some mechanism for production of clinically evident harm can be advanced which provides a plausible reason for expecting linearity.