Interacting with our ever-changing physical environment requires continual recalibration of the motor system. One mechanism for this is motor adaptation. Understanding how motor adaptation is implemented by the human brain, how different regions work in concert to adapt, and how this function relates to metabolic use of neurochemicals poses an important challenge in neuroscience. In humans, motor sequence learning is related to γ-aminobutyric acid (GABA) concentration in the primary motor cortex (M1). However, the role of M1 GABA in adaptation - where behaviour is thought to be acquired outside M1, but retained within M1 - is unclear. In this within-subject, crossover study, we quantified GABA and Glutamate from the hand region of the left human primary motor cortex (M1) using 7T-MR Spectroscopy while participants (n=15) performed a visuomotor task with or without an adaptation component (control condition vs. rotation condition). In the rotation condition, participants were required to adapt their centrifugal shooting movements to a rotation of the visual feedback which increased stepwise by 10 degrees after every block of 40 trials in order to drive adaptation throughout the duration of the scanning session. To probe retention participants performed a washout behavioural task after each MR session. We collected resting-state fMRI data immediately before and after the task. In the rotation condition, participants adapted to the increasing rotation, (LME of error in first and last epoch; effect of epoch χ2(1)=55.58 p<0.01). Participants retained the adaptive movement in the first block of the washout (one-sample t-test of washout error: t = -9.63 p<0.01). We first replicated changes in functional connectivity known to occur in response to adaptation: adaptation increased functional connectivity in a cerebellar network. Further, change in network strength correlated with adaptation (r=-0.65 p=0.01). We next tested our specific hypothesis regarding a link between M1 GABA, M1-Cerebellar connectivity and retention. We found that higher baseline M1 GABA relates to greater retention of adaptation (r =-0.62 p=0.02) but does not relate to adaptation-acquisition (r = 0.07 p =0.82). Moreover, M1-Cerebellar connectivity change is associated with retention (r=0.68, p=0.01), but not adaptation (r=0.05 p=0.87). Finally, M1 GABA relates to M1-Cerebellar connectivity change (r =-0.63 p=0.03). A mediation analysis revealed that M1-Cerebellar connectivity change mediates the relationship between M1 GABA and retention (confidence interval for mediation coefficient excludes zero: ab=-23.87; 95%-CI -62.29, -1.8). Our results showed that a) participants are able to adapt to a stepwise increasing rotation, b) this adaptation process increases connectivity in a cerebellar network and c) retention of the adapted state is associated with baseline M1 GABA. The relationship between M1 GABA and retention is mediated by M1-Cerebellar connectivity change.