During the past two decades, the rodent bioassay for detection of chemical carcinogens has reached a high standard of performance with both an increased number of animals and dose levels and a more detailed assessment of findings. However, the basic principles of testing and evaluation of results have remained essentially unchanged. Problems such as the length of the testing, use of maximum tolerated dose (MTD), selection of strains, variability of spontaneous tumors, discordant results between mouse and rat, and the classification of chemical carcinogens according to their mechanism of action have all remained unsolved. By contrast, the results of short-term tests and of other biological analyses do not always show a direct correlation with those of the long-term bioassays; this can be interpreted as an indication of different mechanisms of carcinogenicity. Currently available medium-term tests may detect carcinogenic activity of chemicals at particular organs in a period of time (weeks to months) relatively shorter than that of the 2-year carcinogenesis bioassay, and they may also provide additional information on mechanisms of carcinogenicity.