Asbestos is classified as a definite human carcinogen, with all forms treated as equally hazardous under current US regulations. Chrysotile is the most commonly used form of asbestos and, while use of chrysotile is restricted in the US, maintenance personnel and building occupants continue to be exposed due to the vast quantities of in-place fiber and use in developing countries is increasing. Despite the accumulated scientific evidence, a number of questions remain unresolved concerning the carcinogencity of asbestos in general and chrysotile in particular. Some researchers argue that the cancer risk associated with exposure to chrysotile is much lower that that for amphibole forms of asbestos. Among workers exposed to chrysotile, an unexplained contrast also exists between the extremely high risk of lung cancer among textile workers and the much lower risk among miners and millers. These controversies are difficult to resolve because only a handful of independent studies have adequate exposure-response information. To address open questions about chrysotile, we propose a historical cohort study of 5818 workers at four North Carolina asbestos textile plants that have extensive data on asbestos exposures but where cancer occurrence has never been studies. The principal goals of this epidemiology study are: 1) To estimate textile workers' exposures to chrysotile asbestos using both established indicators based on fiber concentration and new indicators, which use transmission electron microscopy to account for fiber distribution, dimensions, and surface area; 2) To characterize the quantitative relationship between cumulative exposure to chrysotile and lung cancer mortality while controlling for potentially confounding occupational exposures, and 3) To also evaluate the risk of lung cancer with respect to new indicators of asbestos exposure, including measures of fiber size distribution, fiber dimensions, and fiber surface area. A well-conducted epidemiologic investigations of this new population would provide information badly needed to help answer questions concerning the generalizability of previous study, and about the relative hazardousness if chrysotile. Furthermore, research on lung cancer morality using exposure indices that account more fully for the bivariate distribution of asbestos fiber length and width would provide new data relevant to risk assessments for all forms of asbestos as well as man-made mineral fibers.