Despite increasing standardisation and international harmonisation of the long-term carcinogenicity study in rodents there are still areas of controversy such as high-dose selection, duration of the test period, combining or not combining the chronic toxicity and the carcinogenicity study, the use of historical control data, conditions for redundancy of the study, and significance of studies of restricted design. It is discussed that the 'Maximum Tolerated Dose', or doses above and below the 'metabolic break-point' should be included. In general a test period of 24 months is adequate, but the protocol should be flexible enough to allow extension beyond 24 months. Major advantages of the combined chronic toxicity/carcinogenicity study are that the toxicity and carcinogenicity data are obtained with the same sample of the test compound using the same batch of animals kept on the same diet under the same environmental conditions. It is discussed that the use of historical control data will not lead to a final conclusion on carcinogenic or noncarcinogenic potential of a compound. A long-term carcinogenicity study is considered reduntant (a) when adequate (semi-chronic) toxicity studies including reproduction and mutagenicity tests and pre-screens for carcinogenicity, do not indicate mutagenic or carcinogenic activity, and (b) when there is a wide margin (e.g. 1000) between the 'no-observed-adverse effect level' and the (presumed) exposure level in humans. Studies of restricted design and conduct may clearly demonstrate carcinogenicity but also may easily lead to inconclusive results.