The survival of 132 young adult control beagles and 117 beagles receiving graded injections of 226Ra ranging from 0.27 kBq kg-1 to 384.2 kBq kg-1 body mass was analyzed. The hazards of natural deaths, all deaths in injection groups, and deaths by bone tumors were assumed to follow a Weibull distribution with a common shape factor of 6.3. Only the scale factors of the Weibull distributions depend on the injection level. There were no significant sex differences. The relative risk with respect to controls for all causes of death increases up to 6,925 for 384.2 kBq kg-1. The dependence of the scale factors for all deaths and death by bone tumors on injected activity was fitted to an empirical regression model, which also contains a term representing radiation-caused deaths other than bone tumors. The risk of bone tumors increases nearly as the square of the injected activity, whereas the risk of the other radiation caused deaths increases approximately in proportion to the injected activity. By means of the regression model, it is possible to predict median survival times for all deaths, bone tumor and non-bone tumor deaths for an arbitrary intake level. Also, simple expressions for the fraction of animals with bone tumors and other radiation-caused deaths can be derived. The empirical model of bone tumor induction, which was based on results from a single injection design, can be generalized to an arbitrary systemic intake schema. This is achieved by using the average dose and dose rate to the skeleton as indices of detriment. Applying the generalized model, it was confirmed that no significant differences in survival can be expected for two groups of beagles receiving multiple injections, if compared to the corresponding single injection groups of about the same total activity. The general model also predicts that even extensive protraction of the intake would increase the survival times only to a very limited extent.