Age-adjustment (standardization), the statistical method most used by cancer epidemiologists to express incidence and mortality rates, makes use of an arbitrarily chosen "standard" population to calculate rates that are in reality only abstract index numbers. It is generally known that the "age" of any standard population affects the magnitude of the index, but it is not widely recognized that the use of U.S. Censuses as standards, possibly excepting the 1930 U.S. Census population, distorts the index number. The greater the disparity between the numbers of males and females in middle-aged groups and older, the greater is the distortion. Hence sex as well as age distribution in a standard must be considered if rates are to be expressed in terms of sex. Trends of cancer incidence or mortality will have different slopes depending on the census year chosen as a standard; yet no standard can arbitrarily be called the "correct" one. Age-, race-, and sex-specific rates are more precise and can be made to give considerably more meaningful description of cancer trends, albeit their use requires scrutiny of a greater assortment of numbers.