is not clear. year of observation for the purpose of this calculation because thyroid surgery for the Comparison group was not offered after 1985. have been expected to develop in the Rongelap group by 1963. It is possible, therefore, that the “In this calculation, to obtain the number of absence of nodules in the early years was merely a consequence of random distribution of a relatively person-years of observation the individual agesat the time of the most recent examination were summed, with the exclusion of all years subsequent to thyroid nodule surgery in those cases where it was performed. To use an extreme example, if a 60 year-old person was first enrolled in the Comparison group and examined in 1975 and subsequently never appeared for reexamination, 60 person-years of observation was calculated. The justification for this approach is that it is considered unlikely that any clinically apparent thyroid nodule will spontaneously disappear. In a recent follow-up study of children among whom some nodules had been detected approximately fifteen years previously, only 10 percent of the nodules were no longer palpable (Rallison et al., 1991). Factors influencing data interpretation: It is possible to draw tentative conclusions relevant to issues of radiation injury to the thyroid from data available on the exposed populations alone, thereby avoiding assumptions about the adequacy of a control group. The data underlying the following analyses are shown in Appendix D and grouped andtabulated in Table 3. However, interpretation of the Marshallese thyroid nodule data must be done cautiously because of the small numberof observations that were possible. This is particularly true when the nodules are subgrouped and analyzed by histologic type. In addition, thyroid disease is greatly influenced by gender, thereby further decreasing sample size for some analyses, particularly in males. Another confounding factor, thyroxine suppression, was initiated in 1965 in an attempt to inhibit or prevent the growth of benign and malignant thyroid nodules. This was prescribed only for the exposed Rongelap population, for the risk of nodule development resulting from the much lower Utirik exposure was felt to be small at that time. It therefore becomes difficult in some instances to interpret results in which Rongelap and Utirik data are grouped together. Finally, just why there were no nodules detected during the first nine years of medical team visits (1955-1963) Based on the estimate of nodule incidence of Maxonetal. (1977) two nodules would uncommon abnormality, particularly since the mean age of the Rongelap peopleat the time of exposure was rather low, 27.6 years. Once the first nodules were detected in 1963 it became the procedure of the Marshall Islands Medical Program to include in its medical team a personhighly skilled in thyroid examination, usually an endocrinologist with special expertise in thyroid disease or a thyroid surgeon. This changein procedure introduces a possible bias that is impossible to quantitate. It is possible that some nodules in the exposed population were naturally occurring rather than radiation-induced. However, as there is no way to identify which nodules these were, no attempt has been made tocorrect for their presence. Issues that can be addressed without invoking data from the Comparison group: 1) Is the “epidemic” of thyroid nodules over? The number of patients undergoing thyroid surgery by year over the duration of the Marshall Islands Medical Program is shown in Fig. 3a. The same data are shown in Fig. 3b except that cases are expressed as percent of the population that remained susceptible to new nodule formation; i.e., excluding persons with prior nodules and persons © who had died prior to the year for which a percent was calculated. Clearly the incidence of nodules which began in the mid-1960’s in the exposed Rongelap group has greatly, if not completely, subsided. Their detection spanned 22 years, beginning 9 years after exposure. The nodules in the Utirik group, on the other hand, had a later onset and a later apparent decline. Detection of Utirik nodules has spanned 19 years, beginning 15 years after exposure. In the following discussion the total thyroid absorbed dose in Figures 4-7 represents the acute radiation dose to the thyroid occurring during exposure to fallout prior to evacuation and time to developmentof nodules (years post exposure) refers to the interval in years from exposureto theinitial clinical detection of the nodule. Possibly more appropriate questions to ask are, (1) is the epidemic of adenomatous (nonneoplastic) nodules over, and (2) is the epidemic of neoplastic 17