Is working memory (WM) a unitary construct or is it composed of separable sub-processes with distinct neuroanatomical bases? In line with this latter view, (i) psychiatric and neurological patients often display subtle deficits in specific aspects of WM, and (ii) neurophysiological evidence indicates that WM engages a wide network of regions throughout the brain.To address this question, we designed a WM battery that required participants to maintain the colors and locations of multiple discs over a brief memory delay and report the location of the cued disc. By varying the number of discs to remember, adding irrelevant information during encoding or maintenance, or requiring subjects to reformat or manipulate memory contents, we sought to selectively engage putative subcomponents across trials: storage, selection, resistance, updating, and manipulation.To test the independence of these subcomponents, we collected behavioral data from 200 participants at two timepoints, T1 and T2, ~10 days apart. Individual subcomponent scores were calculated from subjects' behavioral errors. The correlation between scores at T1 and T2 was significantly greater within- relative to between- subcomponents, and data was best described by a model that included at least four subcomponents. These results argue against a unitary WM construct. Additionally, we collected fMRI data from a subset of subjects to determine whether these putative subcomponents engage distinct neural circuits. Representational similarity analysis revealed distinct activity patterns amongst these subcomponents in frontal and parietal cortex. Together, these findings suggest that WM involves selective engagement of multiple subcomponents with distinct patterns of neural activation.