Typical mathematics instruction involves blocked practice across a set of conceptually similar problems. Interleaving, or practice across a set of conceptually dissimilar problems, improves learning and transfer by repeatedly reloading information and increasing discrimination of problem features. Similarly, comparing problems across different contexts highlights relevant and irrelevant knowledge. Our experiment is the first to investigate the relative effects of interleaving geometry problems and interleaving contexts. Thirty-three fourth-grade students received the same practice problems but were randomly assigned to one of three conditions: interleaved by math skill, interleaved by context, and interleaved by math skill and by context (i.e., hyper-interleaved). Afterward, each participant was exposed to tests assessing declarative and procedural knowledge. The results suggest that interleaving math skill within and between varying contexts may enhance the acquisition of mathematical procedures.