It is essential, in order to minimize expensive drug failures, to determine potential toxicity problems as early as possible. In view of the large libraries of compounds now being handled by combinatorial chemistry and high-throughput screening, identification of drug toxicity is advisable even before synthesis. Thus, the use of predictive toxicology is called for. A great number of in silico approaches to toxicity prediction have been described in the literature, but one of the most ambitious goals of QSAR applications to toxicology is modeling of chemical carcinogenicity, which has severe consequences on the quality of life and has led to enormous investments in time, financial resources, and animal lives necessary to test the chemicals adequately. This review attempts to summarize present knowledge related to the computational prediction of carcinogenicity. Several computational protocols are described, ranging from knowledge-based approaches and statistically-based systems to simple and fast procedures based on only the 2-D graphing of the investigated structures. Comparative tests of the ability of these systems to predict carcinogenicity show that improvement is still needed. The consensus approach is recommended, whereby the results from several prediction systems are pooled.