Learning requires applying limited working memory and attentional resources to intrinsic, germane, and extraneous aspects of the learning task. To reduce the especially undesirable extraneous load aspects of learning environments, cognitive load theorists suggest that spatially integrated learning materials should be used instead of spatially separated materials, thereby reducing the split-attention effect (Sweller & Chandler, 1994). Recent work, however, has suggested a new distinction between two common formats of spatially separated displays: spatially distributed versus spatially stacked (Jang & Schunn, 2010). Moreover, a distinction between instructions and learning task materials has rarely been made. Across two studies with 106 college students (56 in Study 1 and 50 in Study 2), we compared spatially distributed (multiple sources of information are placed side by side) versus spatially stacked (only one at the top is visible) instructions, without changing the learning task materials, on both task performance and learning. With materials more typical of practice, Study 1 showed that the distributed-display instructions led learners to more efficient learning; learners finished the task faster and scored higher in the overall learning test. With materials more tightly controlled for spatial format per se, Study 2 replicated the effect and found that the benefit of the distributed instructions appeared to be associated with changes in cognitive load. Implications for educational practice are discussed.