Radiation-induced genomic instability is a modification of the cell genome found in the progeny of irradiated somatic and germ cells but that is not confined on the initial radiation-induced damage and may occur de novo many generations after irradiation. Genomic instability in the germ line does not follow Mendelian segregation and may have unpredictable outcomes in every succeeding generation. This phenomenon, for which there is extensive experimental data and some evidence in human populations exposed to ionising radiation, is not taken into account in health risk assessments. It poses an unknown morbidity/mortality burden. Based on experimental data derived over the last 20 years (up to January 2012) six mechanistic explanations for the phenomenon have been proposed in the peer-reviewed literature. This article compares these hypotheses with the empirical data to test their fitness to explain the phenomenon. As a conclusion, the most convincing explanation of radiation-induced genomic instability attributes it to an irreversible regulatory change in the dynamic interaction network of the cellular gene products, as a response to non-specific molecular damage, thus entailing the rejection of the machine metaphor for the cell in favour of one appropriate to a complex dissipative dynamic system, such as a whirlpool. It is concluded that in order to evaluate the likely morbidity/mortality associated with radiation-induced genomic instability, it will be necessary to study the damage to processes by radiation rather than damage to molecules.