Previous attempts to link leukemia risks to children's home exposures relied upon indirect measures of exposures to carcinogens, based on interviews of households and outdoor levels predicted by Geographic Information Systems (GIS). Such indirect exposure measures can lead to systematic biases in estimated relative risks, associated with particular carcinogens. Also, children's exposures to carcinogens vary randomly within households over time, and thus can introduce attenuation biases into exposure-risk analyses. Given the likely magnitudes of these uncertainties in children's exposures to carcinogens, it is surprising that no studies have characterized systematic and random errors in children's home exposures. Here, we propose a validation/reproducibility study in which we will collect repeated air and dust samples in 250 households selected from a large case-control study of childhood leukemia, namely, the Northern California Childhood Leukemia Study (NCCLS). We will estimate the magnitudes of both systematic errors and random errors and calibrate interview/GIS based exposure surrogates against 'true' contaminant levels measured in air/dust samples. Then, we will use the errors estimated from the validation study to correct naïve estimates of childhood leukemia risk (overall and by subtype) associated with particular carcinogens in the main NCCLS (1,100 cases and 1,650 controls). These analyses will also allow us to identify possible time trends and seasonal changes in exposure levels during critical time windows of gestation and early childhood development. We will focus upon 40 known or suspected carcinogens in house dust and indoor air, including pesticides, persistent organic compounds (polychlorinated biphenyls, polycyclic aromatic hydrocarbons, and nicotine, a surrogate for tobacco smoke), and volatile organic compounds. In each household, a total of three samples of vacuum-cleaner dust and indoor air (using passive air monitors) will be tested for the respective persistent and volatile carcinogens. This proposal presents an exciting opportunity to identify and correct for potential biases of estimated childhood leukemia risks, associated with carcinogen exposures in the NCCLS population. The results would not only clarify the potential importance of particular carcinogens to childhood leukemia risk but also would provide unique information regarding home exposures to carcinogens that can be used in environmental epidemiology and risk management. ExpandCollapse Abstract