The human cerebellum shows fMRI activity during motor, cognitive and social processes, indicating that the cerebellum contributes to each ofthese functions. However, its precise contribution to these diverse domains is not yet clear. One core function of the human brain that involvescerebellar as well as cortical regions is working memory processing. Here, we characterised the cortico-cerebellar working memory network interms of its distinct functional regions and task profiles as a fist step to understanding the cerebellar contribution to working memory. We acquired high resolution (3T) fMRI data from the neocortex and cerebellum of 16 healthy participants performing a working memory task.During the task, participants were required to memorize digits on a screen (encoding phase), which they were later asked to execute with theirright hand (recall phase). Digits were shown sequentially along with a colored box. The box color indicated whether participants would later haveto execute the digit sequence in the same order (forward; yellow box) or in reverse order (backward; blue box). The digit sequence alwaysconsisted of 6 digits, but the number of digits to memorize varied between 2, 4 and 6 digits (memory load). Digits to memorize were highlightedand disappeared after 1s during the encoding phase and were obscured during recall. To determine the cortical working memory regions, we calculated the mean activation map of all task conditions compared to rest. We manuallyparcellated the mean task network into six regions of interest in the left cortical hemisphere: These included superior parietal occipital cortex(SPOC), intraparietal sulcus (IPS), ventral premotor cortex (PMv), dorsal premotor cortex (PMd), V1 and M1. For each region, we derived a taskresponse profile by extracting the mean activation during each of the 12 task conditions. As expected, M1 showed higher activation overall duringthe recall than encoding phase, while V1 showed increasing activation with increasing load size during the encoding, but not retrieval. IPS andPMd showed higher activation during backwards than forwards conditions and higher activation during the retrieval phase compared to theencoding phase. SPOC and PMv showed increasing activation with increasing load size during the encoding phase. To determine the cerebellar regions corresponding to the distinct cortical regions, we used a winner-take-all approach to assign every cerebellarvoxel to a cortical region whose task profile showed the strongest correlation with the task profile of the cerebellar voxel.