Thirty-one sets of data containing lengths and widths of micron- and submicron-sized particles of amphibole asbestos fibers and fragmented amphibole particles and four sets containing summary data are compared in this study. The goals are to describe characteristics of amphibole asbestos fiber populations across asbestos occurrences, to determine which correlate most closely with the generally accepted relative risks for mesothelioma, and to describe the characteristics of amphibole cleavage fragment populations for which an excess risk for mesothelioma has not been established. The analysis shows that modal widths of asbestos fibers vary among geologic occurrences, many asbestos dust populations display multimodal widths, the modes occur at different widths, and the tendency for fibers to disaggregate into component fibrils varies. It also shows that all exposures associated with elevated mesothelioma contain abundant fibers >5 μm in length that have widths <0.25 μm, including fibers <0.15 μm, and that risk for mesothelioma qualitatively corresponds to their abundances. Other parameters that might serve as proxies to correlate with relative risk include fibers with L > 5 μm and W ≤ 0.33 μm and L > 10 μm and W ≤ 0.33 μm. Such particles are absent or occur only in trace amounts in amphibole cleavage fragment populations. Asbestos fibers with such small widths are, for the most part, excluded from occupational exposure assessments which rely on optical microscopy or transmission electron microscopy (TEM) equivalents, compounding the already difficult assessment of risk from inhalation of elongated amphibole particles in a mixed dust environment, especially mining. If we are to understand the outcomes of occupational exposures and the responses observed in animal and cell culture experimentation, the abundance and characteristics of the narrowest elongated mineral particles in the exposure should be established by electron microscopy.