The spatial split-attention principle suggests that presenting related words and pictures spatially close to one another will improve learning compared with a spatially distant design, and two meta-analyses have shown support for the principle. However, it is not clear why the principle occurs. It has been theorized that integrated graphic designs lower the extraneous cognitive load imposed on learners, however, there has not been a systematic review to examine the evidence for this claim. In this study, we systematically review the literature around the spatial split-attention principle in multimedia learning environments to examine if measures of cognitive load support the theoretical prediction. Our analysis of 41 comparisons found that measures of cognitive load largely do not support the extraneous cognitive load explanation, nor did we find compelling evidence that integrated graphic designs significantly influence any measure of cognitive load compared with spatially distant designs. We propose an alternative explanation for the principle in which integrated graphic designs may facilitate the allocation of germane resources and integrative cognitive processing rather than reduce extraneous cognitive load and discuss a theoretical foundation for testing this hypothesis.