Although women have been employed in the nuclear weapons industry since its inception, little is known about the potential health effects that women may experience as a result of work related exposures to ionizing radiation and nonradiation hazards. Studies that have reported results for women have tended to suffer from small numbers of observations, short follow-up, young average age of cohort members and a lack of exposure information. Despite these shortcomings, suggestive but inconsistent elevations for several types of neoplasms have been reported including several known to be associated with ionizing radiation; and for several nonneoplastic conditions. This study attempted to overcome the shortcomings just mentioned by combining cohorts of female nuclear workers from 12 U.S. nuclear weapons facilities. These included: Los Alamos National Laboratory, Zia Company, Rocky Flats, Hanford, Mound, Savannah River, Oak Ridge X- 10, Y - 12 and K-2 5, Fernald, Linde and Pantex. The specific aims of this study were to combine data for female employees from the 12 facilities described above, to estimate doses or exposures to individuals for radiation and . nonradiation hazards, to estimate the relative risk of mortality from neoplastic and nonneoplastic diseases, to estimate the amount of uncertainty associated with these relative risk estimates, and to evaluate the feasibility of conducting nested case-control, case-cohort and morbidity studies among female nuclear workers. The results from this study help to fill a major gap in our knowledge regarding the health of female nuclear workers. In collaboration with researchers at the University of North Carolina, we developed questionnaires on radiation dosimetry practices and data resources, and on physico- chemical exposures, industrial hygiene practices and data resources. The radiation dosimetry questionnaire expanded on a questionnaire that had been previously devised by staff at the Department of Energy. These questionnaires were sent to designated contacts at the study facilities. Unfortunately, fewer than half of the questionnaires were completed and returned. When mortality for the combined cohort is compared with U.S. death rates, fewer deaths than expected are observed for most causes of deaths. Exceptions are deaths from mental disorders (Standardized Mortality Ratio (SMR) =147), certain genito-urinary system diseases (SMR= 129), as well as symptoms and ill-defined conditions (SMR=163). Mortality from conditions that have in the past been found to be associated with exposures to ionizing radiation is not higher than expected, or was close to expectation. A strong healthy worker effect is observed for the entire cohort and for each individual subcohort with the exception of Linde, in which case the observed number of deaths is similar to the number expected. The weaker healthy worker effect observed among Linde workers is largely due to more deaths than expected from ischemic heart disease. The healthy worker effect is observed among workers who were monitored for external radiation exposures and among workers who were not monitored for external radiation exposures. The SMR (observed/expected deaths X 100) for all causes of death combined is 78 for unbadged and 69 for badged workers. More observed than expected deaths among both monitored and unmonitored women are evinced for mental disorders. Increased SMRs are observed among unmonitored employees for deaths from symptoms and ill defined conditions, diseases of the genito-urinary system and for homicide. Among badged workers, deaths from ill-defined conditions is as expected, and lower than expected for diseases of the genito-urinary system and homicide. When time dependent proportional hazards analyses of cumulative penetrating doses for all monitored employees, regardless of length of employment, are performed for all facilities combined, the relative risk of death per rem increases with increasing cumulative penetrating dose for all leukemias combined (ICD8: 204.0-207.9), other than chronic lymphatic leukemia (ICD8: 204.1 (RR/rem=I.13, 95%CI=I.02-1.25". Relative risk estimates per rem are suggestively elevated for all cancers combined (ICD8: 140239.9 (RR=1.03, 95%CI=0.99-1.06", breast cancer (ICD8: 174.0-174.9 (RR/rem=1 .05, 95%CI=0.99-1.12" and for all hematologic cancers combined (ICD8=200.0-209 . (RR/rem--l.08, 95%CI=0.99-1.17". Relative risk estimates do not increase per rem for any of the other groups of causes of death that we investigated (radiosensitive solid tumors: 150.0-150.9,151.0-151.9,153.0-153.9, 162.0-162.9, 174.0-174.9, 188.0-188.9, 189.0-189.9, 191.0-191.9, 192.0192.9, lung cancers: 162.1, ovarian cancers: 183.0-183.9, brain cancers: 191.0-192.9,225.0-225.9, and 238.1-238.9, thyroid cancers: 193.0-193.9). When data from individual facilities are analyzed, increased effect estimates from all cancers (RR/rem = 1.131) and from radiosensitive solid tumors (RR/rem = 1.16) are observed at Savannah River, and increased relative risks for hematologic cancers (RR/rem = 1.25) and for leukemia (RR/rem = 1.32) are observed at X-I O. Inclusion of lag times, or length of employment in these models does not change the results. These results should be interpreted with caution. Additional research is needed to evaluate the impact of potential confounders that we have been unable to account for in this study. These include potential confounders such as socio-economic status, smoking and other life style activities, time related factors, potential errors in radiation dosimetry, the influence of other work site exposures and other factors. Reliance on mortality data raises concerns especially regarding the increased occurrence of mortality observed for mental disorders. Although one may hypothesize stress related illness as a possible explanation, further evaluation of the specific diagnoses comprising this combined category of mental disorders is first required. Finally, the evaluation of associations between cause-specific mortality and cumulative doses suffers from a relatively small number of deaths, doses that are skewed toward the very low doses and few observations at higher doses.