Category learning is often modeled as either an exemplar-based or a rule-based process. Variations on the EBRW exemplar model of Nosofsky and Palmeri (1997) and the RULEX rule-based model of Nososfky, Plameri, and McKinley (1994) were implemented in the ACT-R cognitive architectures. The architecture allows the two classifications to be merged to produce classification behavior. The merged system reproduces latency, learning, generalization, and response consistency data from three category learning experiments - Nosofsky and Palmeri (1997), Nosofsky, Palmeri, and McKinley (1994), and Erickson and Krushke (1998).