Descriptive statistics in the field of medicine are usually based on univariate analysis. However, a multivariate descriptive analysis can often be usefull to jointly describe all variables considered for study. This multivariate description is difficult to perform and visualize for more than three variables at a time. Multiple correspondence analysis (MCA) provides a means of performing multivariate description of categorical data. The method consists in projecting the data of an n-dimensional space which is constituted by the variables under study onto a succession of two-dimensional planes. The relationships between variables can then be deduced from the relative positions of the modalities of the variables on the planes. At the same time, numerical indices are used in parallel to specify and validate the observed relationships. The use of MCA is illustrated with a prospective series of renal carcinomas for which different histological characteristics are given. The main applications of MCA are detailed with comments on practical implementation.