[en] Human data are the most convincing data--even when their applicability may be considerably strained--and the acquisition and interpretation of such data deserves a high priority. This would include currently funded studies on plutonium workers, thorium workers, and uranium diffusion plant workers. It is obvious, however, that there is also need for information totally unavailable from human sources. Appropriate experimental animal studies must provide this information. These studies must be designed to permit the most credible extrapolation to man, and the development of such extrapolation techniques is as important as the conduct of the studies themselves. There was agreement that the results of toxicity experiments are almost never reported in a manner most useful to modelers and that this situation should be rectified by some means short of dictatorial edict: perhaps a manual of recommended practices, perhaps a ''flying squad'' of consulting epidemiologists, perhaps more frequent workshops involving the people who actually gather and report the data. Among the forbidden practices would be the expression of radiation dose in rem and the reporting of tumor incidence without individual animal data on time at risk. In the opinion of the panel, a delay in the implementation of plutonium recycle or the LMFBR increases, rather than decreases, the need for additional experimental data on pluonium toxicity. Without the delay there might be insufficient time for the acquisition and effective application of data from newly initiated research programs. But when the inevitability of plutonium as an energy source is acknowledged--5, 10, or 20 years hence--we will be severely and justifiably criticized if we have not used this period to acquire a better understanding of the health effects of plutonium