Mortality rates for all malignant neoplasms (combined) and for 34 site-specific cancer categories for selected high and low altitude populations are compared using two related techniques: confidence interval overlap for strictly descriptive purposes and analysis of standardized mortality ratios using lower and upper 95% statistical significance factors for the ratio of an observed value of a poisson variable to its expectation. Techniques are employed to minimize confounding due to industrialization, urbanization or selected cultural characteristics. Cancer mortality data for U.S. counties averaged over the 20-year period, 1950-1969, were used. For most comparisons a deficit in cancer mortality in high altitude counties was observed. The largest differences between the low and high altitude groups were found for cancers of the tongue and mouth, esophagus, larynx, lung and melanoma. Some limitations of ecologic studies are discussed. © 1982 IRL Press Limited.