The cognitive theory of multimedia learning postulates learning information in a dual-modality design is more effective than in a single modality, which is known as the modality effect. Research has found that the modality effect supports problem-solving learning, but not retention-based learning. This divergence in findings can be explained by the mediating effects of individual differences in visuospatial working memory. The current study confirmed the modality effect, showing superiority of dual-modality over single modality. Results revealed a significant interaction between visuospatial working memory and modality by retention, but no such interaction was obtained for problem solving. It was found that learners with low visuospatial working memory scored higher on retention in dual modality than these in single modality. Interestingly, no differences were found for learners with high visuospatial working memory on retention in both single- and dual-modality conditions. The findings may explain for whom the modality effect works in relation to retention. Finally, the analysis showed no association between modality conditions and subtypes of cognitive load, that is, extraneous and germane cognitive load.