General recognition theory (GRT) is both a theory of categorization and a framework for studying human categorization behavior. The GRT toolbox is a set of MATLAB scripts and subroutines that can help an experimenter design categorization experiments, generate stimuli for these experiments, simulate a participant’s responses, analyze categorization data, and graph results. The typical user designs experiments for two-category tasks in which the categories are specified by multivariate normal distributions. The toolbox also provides tools for fitting the general linear classifier and the general quadratic classifier to a data set.