cesk Agate governments. Conversely, if the diets of the people at Likiep, Ailuk, Wotho, and Mejit were to include more imported foods, they would be more like the BNL community B or MLSC diets, in which case the doses would be much lower than listed here. Even if the use of imported and local foods remains as it currently is, there is a possibility that the average intake of local foods could be greater than we have assumed in our model diets--for example, if the entire BNL diet rather than the MLSCresults were assumed to apply to Ujelang Atoll. The reasons for our selection of the dietary intakes used here are discussed above in Limitations of the Assessment. Third, the range of values observed for the retention of 1376, in humans has been summarized by the ICRP 36137 and the NCRP.?% For example, the range of observed values for the retention time for the short-term compartment is 0.5 to 2.1 d with a mean of | d; the upper limit that has been observed is greater than the mean by only a factor of 2. For the long-term compartment, the data range from 60 to 165 d with a mean value of 110 ds; the maximum value in this case is less than twice the mean value. The fraction of the intake that has been observed to go to the short-term compartment (ise., 2-d) ranges from 0.02 to 0.22 with a mean of 0.1; for the long-term compartment(i.e., 110-d), the range is 0.78 to 0.97 with a mean value of 0.9. For both cases the maximum value is less than twice the mean. Previous evaluations indicate that dietary intake in a population is log-normally distributed. Our evaluation of the MLSC survey confirms this log-normal distribution. The distribution of doses is also log normal and the mean dose calculated using the average value for all model parameters falls at about the 68th percentile; that is, 68% of the population would be expected to have a dose at or below the listed mean value. A dose equal to twice the mean value will include 88% of the population. It is important to recognize when we talk about the average doses here that they are not at the midpoint (or 50% point) of the distribution. There are several reasons why the average doses we present might be lower. (1) The doses are calculated assuming residence since 1978. For uninhabited atolls, doses would be expected to be about 2.3% lower per year until resettlement occurs based on the radiological decay of cesium and strontium. (2) We still do not know the environmental! residence time of cesium in the atoll ecosystem. If it were 30y (i.e., equal to the radiological half life), the estimated doses would be half (50%) of those presented in the tables. If the environmental residence time were as long as 50 y, the doses would be 34% 53