A large collection of histology slides from autopsies of some 29,000 former residents in uranium mining districts of east Germany is currently being catalogued and studied at the German Center for Cancer Research in Heidelberg, Germany. We have explored systematically and in detail whether and how we might gain access to this material, and whether we could use it to test certain specific hypotheses about how exposure to silica dust increases risks of lung cancer. The underlying idea is that silica dusts, or their fibrotic effects in the lung and lymph nodes, might interact biologically with other carcinogenic agents, including ionizing radiation, in a way that potentiates lung cancer risks. The postulated bio-mechanisms would be expected to result in unusual distributions of lung cancer cell types. Those distributions of cell types can be identified only by appropriately designed histo-pathologic (rather than death-certification-based epidemiologic) studies. With the collaboration of the German authorities responsible for maintenance of the materials and data concerned, we identified a sample of 302 former east German uranium miners included in the autopsy archive. The sample was selected from all those potentially available so as to generate as wide a range as possible of exposures to silica dust and to ionizing radiation. Estimates of cumulative exposures to silica dust, to ionizing radiation, to asbestos, and to arsenic were obtained by arranging for computerization of very detailed manuscript occupational histories of the individuals sampled, and by integrating these data with assessments, by three individuals familiar with underground conditions at the mines over the years, of mine-, job- and calendar-time-specific concentrations of the occupational pollutants. More than 2,700 histology slides, referring to 292 of the 302 miners sampled, were retrieved from the archive and were examined independently by four pathologists in the USA. The laboratory examinations were made according to a protocol based on World Health Organization criteria for lung cancer typing. The distribution of time worked in uranium mines of those sampled ranged from a few weeks to 41 y (median 15.7 y). Estimates of cumulative exposures to silica and to radon daughters were correlated but not collinear (r = 0.73). Correlations between average intensity of exposure to radiation (WLM per exposure year) and cumulative exposures to arsenic, to asbestos, and to silica were low (r = 0.20, -0.20 and 0.23 respectively.) The patterns of joint distributions of occupational exposures suggest that appropriate further sampling from the autopsy archive can provide data that could be used to test the underlying research hypothesis - provided that adequate histologic material is available. Information on smoking habits was obtained for only 52% of those sampled; most (92%) of them had smoked at one time or another. It is unlikely that more complete data on the smoking habits of the miners of interest can be established from the available records. All four pathologists were able to distinguish between different pathological types of primary lung cancer in about two thirds of the sampled cases. Concordance between any two of the pathologists in their designations of five lung cancer cell types ranged from 60% to 76%. Most (93%) of the sets of slides included some presenting lymph nodes. About 86% of the cases provided sets of slides that were interpretable regarding the presence or absence of fibrosis at different anatomical sites. The quality of these slides was sufficient for all four pathologists to distinguish between cases that indicated the presence of lymph node silicosis only, parenchymal silicosis only, and the presence of both types of silicosis; but there were some marked differences between the pathologists about which particular cases exhibited the features concerned. Pre-neoplastic lesions were identified on slides for between 10% and 20% of the cases. Levels of agreement between the pathologists regarding which cases were affected was low, ranging from 13% to 32% of cases where either one of two of them recorded the presence of a pre-neoplastic lesion. Access to histologic material that would be required for the research envisaged has been verified in practice and is assured for the future. Laboratory examinations of the material have demonstrated that it can be used for the kind of studies required, but further discussion and standardization of pathologic criteria are necessary before continuing with the work. Data on occupational exposures of the uranium miners considered, to silica dust and to various occupational carcinogens, are retrievable. The sampling strategy adopted for the feasibility study has succeeded in generating distributions of estimates of exposures that, in principle, would permit rigorous testing of the main research hypothesis. Some adjustment to that strategy should be considered, to increase the proportion of subjects for whom smoking data are likely to be available. The results indicate that a further sample of about 600 individuals from the autopsy archive would provide sufficient statistical power (> 80%) to discriminate, at least at a 5% significance level, between the underlying research hypothesis and its negation. We recommend that such studies should be pursued.