One great obstacle to understanding and using the information contaned in the genotoxicity and carcinogenicity databaes is the very size ofsuch datalses hir vastness mnkes them difficukt to read; this leads to inadequateptation of the information, which becomes costly in terms oftime, labor, and money. In its search for adequate approaches to the problem, the scientific community has, curiously, almost entirely neglected an existent series of very powerful methods of data analysis: the multivariate data analysis techniques. These methods were specifcally deigned for exploring large data sets. This paper presents the multivariate techniques and reports a number ofapplications to genotoxkcty problems.These studies show how biology and mathematical modeling can be combined and how successfil this combination is.

No abstract