Eddy sand bars are a major temporary storage site for the suspended load of the Colorado River through Grand Canyon. Recent research has demonstrated that there is large temporal and spatial variability in the size of these bars. Such large variability calls into question the accuracy of past assessments of widespread erosion of sand bars in Grand Canyon, hinders attempts to define longitudinal trends in bar size, and presents a challenge in the development of monitoring programs of the on-going effects of dam operations. Synthesis of oblique photographs taken during the past century, aerial photographs during the past 70 years, and topographic surveys during the past 30 years, demonstrate that site-to-site variability in bar response to flow and sediment-transport regime decreases if one only considers the upper parts of these bars. However, there is large site-to-site variability if one compares the size of the entire mass of sand stored in each eddy within any river reach. Upper parts of eddy bars are formed by discrete floods, are deflated by wind during intervening periods, and reflect the temporal sequence of large magnitude flow events. The size of the lower parts of eddy bars are highly variable from site to site and reflect both large-scale characteristics such as system-wide sand storage and small-scale processes such as subaqueous mass failures. Historical records of bar change that depict sediment storage of the entire area of subaerial sand exposed at base flow do not indicate significant post-Glen Canyon Dam changes. However, the record of bar change of the high-elevation parts of bars indicates system-wide reduction in the area of the upper parts of bars and increasing downstream deposition of high-elevation bars during discrete managed floods. This longitudinal trend occurs because the suspended load increases in the downstream direction.