ABSTRACT Background The modality and redundancy principles are well‐established within the cognitive theory of multimedia learning for their effectiveness in traditional media. However, previous studies applying these principles in virtual reality have shown contradictory results when immersive technologies are involved. Objectives This study investigates the effects of modality and redundancy principles from the cognitive theory of multimedia learning on cognitive load and learning outcomes in head‐mounted display augmented reality. Methods Using a between‐subjects experimental design, 104 male participants were randomly assigned to three conditions based on the way verbal instructions related to a T‐shirt folding procedure were presented (audio‐only, text‐only, text‐and‐audio). The presentation of the verbal instructions served as the manipulated variable; cognitive load levels and participants' performance were measured as outcome variables. Bayesian analyses were conducted to verify that there were no significant differences between the text‐and‐audio group and the audio‐only group, and a Partial Least Squares Structural Equation Model (PLS‐SEM) examined the relationship between the experimental condition, cognitive load levels, and participants' performance. Results and Conclusions Results support the validity of the modality principle in head‐mounted display augmented reality, showing that applying the modality principle significantly decreases intrinsic cognitive load and improves learning performances. The null hypothesis was supported by comparing audio‐only and text‐and‐audio conditions for both cognitive load and learning outcomes. These findings suggest that cognitive theory of multimedia learning principles work similarly in head‐mounted display augmented reality as in traditional media, and they offer valuable insights into multimedia learning in immersive augmented reality, extending our understanding of how these principles function in immersive technologies. Contradicting results using immersive technologies could be explained by considering task complexity and perceived intrinsic cognitive load; however, further studies testing different learning materials are necessary to support this hypothesis.