This article provides a tutorial overview of cognitive architectures that can form a theoretical foundation for designing multimedia instruction. Cognitive architectures include a description of memory stores, memory codes, and cognitive operations. Architectures that are relevant to multimedia learning include Paivio's dual coding theory, Baddeley's working memory model, Engelkamp's multimodal theory, Sweller's cognitive load theory, Mayer's multimedia learning theory, and Nathan's ANIMATE theory. The discussion emphasizes the interplay between traditional research studies and instructional applications of this research for increasing recall, reducing interference, minimizing cognitive load, and enhancing understanding. Tentative conclusions are that (a) there is general agreement among the different architectures, which differ in focus; (b) learners' integration of multiple codes is underspecified in the models; (c) animated instruction is not required when mental simulations are sufficient; (d) actions must be meaningful to be successful; and (e) multimodal instruction is superior to targeting modality-specific individual differences.