Previous research has found that adding pictures to either text-based learning- or testing-materials has beneficial effects (i.e., multimedia effects in learning and testing). However, a potential interaction of multimedia effects in learning and in testing has scarcely been the focus of research so far. Accordingly, in the present experiment, N = 129 university students received text-based material that was either complemented by pictures both in learning and testing (P-P), only in learning and not in testing (P-T), in testing but not in learning (T-P), or not at all (T-T). Students performed better in the P-P condition than in the other three conditions (P-T, T-P, T-T). Differences between the latter three conditions were not significant. Thus, pictures needed to be present in both learning and testing to foster students' performance. Nonetheless, pictures boosted students’ confidence even if performance was not actually improved, thereby underlining the risk of picture-induced metacognitive bias. •We investigated interactions of multimedia effects in learning and testing.•Pictures in learning were only helpful when pictures were also available in testing.•Pictures boosted students' confidence in their performance (multimedia heuristic).•Using multimedia incurred the risk of inflating metacognitive judgments.