Studies have indicated that pictures in test items can impact item-solving performance, information processing (e.g., time on task) and metacognition as well as test-taking affect and motivation. The present review aims to better organize the existing and somewhat scattered research on multimedia effects in testing and problem solving while considering several potential moderators. We conducted a systematic literature search with liberal study inclusion criteria to cover the still young research field as broadly as possible. Due to the complexity and heterogeneity of the relevant studies, we present empirical findings in a narrative review style. Included studies were classified by four categories, coding the moderating function of the pictures investigated. The evaluation of 62 studies allowed for some tentative main conclusions: Decorative pictures did not appear to have a meaningful effect on test-taker performance, time on task, test-taking affect, and metacognition. Both representational and organizational pictures tended to increase performance. Representational pictures further seem to enhance test-taker enjoyment and response certainty. Regarding the contradictory effects of informational pictures on performance and time on task that we found across studies, more differentiated research is needed. Conclusions on other potential moderators at the item-level and test-taker level were often not possible due to the sparse data available. Future research should therefore increasingly incorporate potential moderators into experimental designs. Finally, we propose a simplification and extension of the functional picture taxonomy in multimedia testing, resulting in a simple hierarchical approach that incorporates several additional aspects for picture classification beyond its function.