Realistic visualizations are considered to introduce the risk of distracting learners from relevant information. In two experiments, the interplay between realism and a known form of distraction, the split-attention effect, were investigated. This effect describes that spatially separating relevant information can have a substantial negative effect on learning. The experiments were conducted using short anatomy learning tasks to test whether a combination of realism and split attention would lead to the worst retention performance or, alternatively, whether realism can counteract the negative effects of split attention. The first experiment (n = 125) revealed that realism attenuated the cognitive load induced by split attention, suggesting a compensatory effect of realism (i.e., realism may have helped learners to deal with the detrimental influence of split attention). However, retention performance was not impacted in a similar way, indicating that this compensatory effect on subjective cognitive load may actually be the result of learners’ illusion that realistic details are helpful. Split attention significantly reduced retention performance. Experiment 2 (n = 152) resulted in negative effects of realism and split attention on retention. In sum, the experiments suggest that realistic details can affect learners independently of other visual design factors as exemplified by the split-attention effect. Thus, the assumption that realism is likely to distract learners is rendered implausible by the experiments, as the distraction of split attention should have amplified any distractive potential of realistic details. However, the results also suggest that the effects of realism on learning are still somewhat unpredictable.