This study investigated how best to study a matrix. Fifty-three participants studied a matrix topically (1 column at a time), categorically (1 row at a time), or in a unified way (all at once). Results revealed that categorical and unified study produced higher: (a) performance on relationship and fact tests, (b) study material satisfaction, and (c) associative strategy use than topical study. A supplemental study examined the benefits of adding signals to the unified matrix. Results showed that signaling produced greater use of global associative strategies and greater learning of global relationships. Findings were explained with respect to cognitive load theory. Implications for studying matrices were as follows: (1) do not just study a matrix topically, (2) study a matrix categorically, and (3) study a matrix in a unified way, but do so in a way that fosters comparative associations across multiple topics and categories. (Contains 3 tables and 1 figure.)