see the blast, some misclassification of exposure status is possible, since they would not have had the personal experience of this dramatic event to date their 1954 atoll of residence. However, if such persons incorrectly recalled their location in 1954, the error would have probably been another southern atoll, since they would likely have remembered the BRAVOtest had they lived on a northern atoll. In addition, transportation to and from theseislands in the 1950s was not frequent, so the likelihood of misclassifying exposure remains small. Theissue ofmultiple exposuresarises in this study population since there were 66 announced nuclear tests in the Marshall Islands between 1946 and 1958. Many of these tests,“ however, took place on Eniwetok Atoll, which is located about 200 miles west of Bikini. In addition, most of these tests were conducted when the prevailing winds were heading away from the Marshall Islands. More importantly, the BRAVO test was the largest of the 66 nuclear tests; it is the only test that people on distant atolls recall having seen. Thus, while it is possible that atolls close to Bikini, such as Rongelap, may have been exposed on multiple occasions,itis unlikely that such exposure occurred on distant atolls. Because this study was a retrospective cohort design, the important issue of latency cannot be addressed. Prospective studies of the Rongelap and Utrik populations reported a mean latency for thyroid nodules of 13 years for Rongelap children exposed at ages less than 10 years. The Utrik children, with lower thyroid doses, had a mean latency of 25 years. Whether persons exposed to smalier doses in the present study may exhibit even longer latent periods is unknown.Since latent periodsat least as long as 34 years are thoughtto exist in other populations exposed to thyroid irradiation,” it will be necessary to continue close follow-up of this population. Theresults of this study suggest that the northern atolls used in previous studies as a source for unexposed controls, with a prevalence of nodules of 6.3%, were inappropriately selected, since the prevalence in our study continues to decrease to less than 1% for the southern atolls, which are located the farthest from the Bikini test site. We believe that a better estimate of the prevalence of thyroid nodules in unexposed Marshallese is 2.45%, the mean prevalence of the two southernmost atolls. Since the prevalence continues to decrease to a value less than 1% for the atoll farthest from the blast site, 2.45% is probably a conservative estimate for the spontaneous or backgroundrate of solitary thyroid nodules in the Marshall Islands. Because authors of previous studies used the prevalence of 6.3% for presumably unexposed controls, their risk co- efficient of 830 excess cases/Gy/y/1 x 10° persons (8.3 excess cases/rad/y/1 million persons)” underestimatesthe truerisk. Using our estimate of 2.45% for the prevalence of nodules in unexposed Marshallese, our new risk coefficient is 1100 excess cases/Gy/y/1x10° persons (11.0 excess cases/rad/y/1 million persons). This is 33% higher than the previous estimate and is quite close to a published composite estimate of 12.3 (1230) for gamma radiation.” The components of radiation dose to the thyroid gland in Marshall Islanders exposedto fallout are relatively unique amongstudies of humansin whom thyroid neoplasia has developed from ionizing radiation. While gamma radiation accounts for part of the total thyroid dose in the Marshall Islands exposure (4% to 16%), the majority of the thyroid dose came from the short-lived radioiodines, “I, “I, and ™I, and, to a lesser extent, "I." There islittle information in the literature, other than that from exposures in the Marshall Islands, concerning the effects of these radioiodines in humans. Although “I alone is known to induce thyroid neoplasms in animal studies,“ it is much less effective in the induction of human thyroid neoplasms, possibly 50 times less so than gammairradiation.” Indeed, studies of *1I therapy in Graves’ disease have led to doubts about whether “I alone induces thyroid nodules in humans.***?” One explanation for the ineffectiveness of “I as a carcinogen in these studies may be that autoimmune thyroid disease renders the thyroid glandresistant to the developmentof neoplasms from “T irradiation. An additional factor is the nonuniform distribution of “I within thyroid tissue compared with gammaradiation; the dose from this type of distribution can ablate tissue at localized “hot spots” and result in a lower dose to the remaining thyroid tissue. Other explanations include the lower dose rate of beta-emitting I compared with gamma radiation and the decreased potential of the thyroid gland to undergo malignant transformation once ablative doses of “I have been received by the entire gland.® However, while the role of “I as an inducer of thyroid neoplasia remains controversial, it should be emphasized that radioiodine fallout contains not only “I but a mixture of short-lived, higher-energy radioiodines. The public health implications of these results are important not only to the Marshallese people but also to populations that may be exposed to shortlived radioiodines from fallout such as may occur during nuclear reactor accidents. These isotopes include the higher energy beta-emitters I, “I, and “I and do appear to be effective inducers of thyroid nodules. In our study, we found the absolute risk coefficient to be nearly identical to the estimate for gammairradiation. Thus, populations exposed to radioiodine fallout should not only be considered for potassium iodide prophylaxis at the time of contamination but should also be carefully followed up for the late development of thyroid nodules. We anticipate the expected rates of such neoplasms to be similar to those found from gammaradiation. The field work for this study was supported by the Marshall Islands Atomic Testing Litigation Project, Los Angeles. The analysis and preparation of this manuscript was supported in part by a grant from the Robert Wood Johnson Foundation, Princeton, NJ. We are indebted to the following Marshallese field staff: Staff Director Atra Lang, Leilani Lokboj, Julie Lloyd, Winnie MacQuinn, Aida Nashion, Anibar Timothy, and Lijon Eknilang. Special gratitude is expressed for the cooperation of the people of the Marshall Islands, the loca] council governments on the atolls, Minister of Health Dr Jeton Anjain (1982), representatives of the Republic of the Marshall Islands, and to all members of the Mar- shall Islands Atomic Testing Litigation Project. We are also indebted to Robert Griep, MD, and Lori Bernstein for participation in the validation aspects of this study; to Bruce Psaty, MD, MPH, Tom Koepsell, MD, MPH, and Linda Rosenstock, MD, MPH, for critical review of the manuscript; to Mary Miller for assistance in data abstraction; and to Paulette Gilliam for preparation of the manuscript. | References 1, Conard RA, Rall JE, Sutow WW:Thyroid nodules as a late sequela of radioactive fallout, in a Marshall Island population exposed in 1954. N Engi J Med 1966;274:1391-1399, 2. Conard RA, Paglia DE, Larsen PR, et al: Review of Medical Findings in a Marshallese Pomulation 26 Years After Accidental Exposure to Radioactive Fallout, US Dept of Energy publica- JAMA,Aug 7, 1987—Vol 258, No. 5 tion (BNL) 51-261 Upton, NY, Brookhaven Na- tional Laboratory, 1980. 3. Conard RA, Knudsen KD, Dobyns BM, et al: A 20-Year Review of Medical Findings in a MarShallese Population Accidentally Exposed to Radicactive Faiiout, US Dept of Energy publication (BNL) 50-424. Upton, NY, Brookhaven National Laboratory, 1975. 4. Conard RA, Sutow WW, Lowrey A, et al: Medical Survey of the People of Rongelap and Utrik Islands 13, 44 and 15 Years After Exposure to Fallout Radiation, US Dept of Energy publication (BNL) 50-220. Upton, NY, Brookhaven National Laboratory, 1970. 5. Conard RA, Meyer LM, Sutow WW,et al: Medical Survey of the People of Rongelap and Thyroid Neoplasia—Hamitton et al 635