The modality effect occurs when people learn better from a combination of pictures and narration than from a combination of pictures and written text. Despite the strong empirical results in earlier studies, the modality effect has been less prominent in later studies of children in learner-paced settings. However, the generalizability of these results in practice may be limited because the studies included notable differences compared to a classroom context. The present study examined the modality effect in a learner-paced classroom context. In a within-subjects experiment, fifth graders learned from illustrated texts and videos and completed pre-, post-, and delayed tests on two science topics. The video group outperformed the illustrated text group in retention, delayed retention, cognitive load, and efficiency measures but there were no statistical differences in transfer. In both learning conditions, the cognitive load was moderate and did not correlate with any learning outcomes. The results suggest that while the modality effect can occur in a learner-paced classroom context, it may not be based on the avoidance of cognitive overload. Alternative explanations concerning the differences in settings and materials between classroom contexts and modality effect research are discussed.