Ionizing radiation has been known as a cause of most forms of leukemia for over 50 years (1), and quantitative models exist for the association between leukemia and external ionizing radiation [other than chronic lymphocytic leukemia (CLL)], based primarily on studies of populations exposed to relatively high doses (2-4). Quantitative estimates of risk at low cumulative doses (e.g., below 100 mGy) are still uncertain, but they have recently been estimated among a combined international study of nuclear workers to be similar to results seen among those exposed to high and moderate radiation doses (5). CLL is the only cancer specifically excluded from compensation programs in the U.S., since it is presumed to be a non-radiogenic cancer. The purpose of this study was to assess quantitatively the association between radiation from all workplace sources and leukemia, controlling for known or suspected confounders of the association. A nested casecontrol study was conducted among workers at four DOE facilities (Hanford, Los Alamos National Laboratory, including Zia workers, Oak Ridge National Laboratory, Savannah River Site) and the Portsmouth Naval Shipyard, sites thought to have minimal exposure potential to plutonium and other internal emitters. Mortality risks were analyzed separately for CLL and non-CLL leukemia, given the expectation of differential radiogenicity for these two outcomes. Leukemia cases were ascertained by death certificate diagnoses, and leukemia subtype was further identified using available medical records. Forty-three cases of underlying CLL deaths and 172 age-matched controls, and 206 non-CLL cases and 823 age-matched controls were identified with follow-up through 1990 to 1996. Bone marrow doses were estimated for each worker from both external sources and plutonium (6), based on monitoring records. Recorded doses were adjusted for measurement biases from exposure to heterogeneous radiation fields, calibration methods, dosimeter design, dosimeter energy response, and geometry of the critical organ (7). Individual external doses from y- and X-ray exposures below the detection limit were estimated (7-9), as were doses from work-related medical X-ray examinations (10). Potentially confounding exposures included cigarette smoking (estimated using workplace medical records) and benzene and carbon tetrachloride exposures. Exposures to the solvents were estimated through a job-exposure matrix developed using workplace records combined with facility process descriptions. Conditional logistic regression was used to estimate the excess relative risk (ERR) per unit of radiation, considering potential confounders such as sex, race and ethnicity, birth cohort, smoking and solvent exposure. Dose lags of 2 years for non-CLL leukemia and 10 years for CLL were employed. For non-CLL leukemia, the ERR per 10 mSv was 1.44% (95% CI: <- 1.03%, 7.59%), adjusting for age, sex and benzene exposure, but it was higher for workers born after 1921 compared to workers born earlier or when excluding leukemias of uncertain subtype. This increased risk was not restricted to high-dose workers: Excluding workers receiving >100 mSv, the sex- and benzene-adjusted ERR per 10 mSv was 6.82% (95% CI: -2.87%, 24.1%). The results suggest that overall risks per unit dose among these nuclear workers are comparable to those observed in high-dose populations, although no evidence was observed of a linearquadratic dose response. For CLL, the ERR per 10 mSv was -2.0% (95% CI: <0, 14%) based on all worker exposures. However, excluding high-dose workers substantially changed these estimates: For workers receiving <100 mSv, the ERR per 10 mSv was 15% (-4.7%, 76%). Time window analyses among this group suggested that a latency of 10 years is appropriate. Adjustments for other suspected confounders had little effect on these estimates. Results for non-CLL leukemia analyses are described in detail elsewhere (11). This large study is among the first to specifically evaluate benzene risk in conjunction with ionizing radiation. Since workers with higher benzene exposure scores had higher rates of non-CLL leukemia than unexposed workers, the elevated risks observed among workers with joint exposure to radiation and benzene are also of interest. Studies based on populations with high radiation doses or low background rates of CLL have observed no effect of radiation on CLL risk, leading to the conclusion that CLL is non-radiogenic. However, this study of CLL in a nuclear workforce showed inconsistent evidence of risk associated with low-level exposures: Although confidence intervals are wide, elevations in estimated risk observed below 100 mSv were not seen at higher exposures. Further studies of larger, low- or moderate-dose populations with substantial background risk and lengthy follow-up may be needed to further address the radiogenicity of CLL. Disclaimer: The findings and conclusions in this presentation have not been formally disseminated by NIOSH and should not be construed to represent any agency determination or policy.