To examine the role of cognition in device adoption, skill retention, and withdrawal from a study of an advanced upper limb prosthesis (the DEKA Arm).T-tests and Wilcoxon rank-sum tests were used to compare test performance among study completers and non-completers. Multivariable regression analyses were used to predict study withdrawal and DEKA Arm skill retention.Compared to self-withdrawn participants, those who were withdrawn by study staff performed significantly worse on tests indexing processing speed, set-shifting, and memory encoding. DEKA Arm configuration (transradial, transhumeral, shoulder - based on amputation level), was a stronger predictor of skill retention than neuropsychological test performance.Frontally-mediated cognitive skills may influence the successful adoption of the DEKA Arm. DEKA arm configurations at higher amputation levels (e.g., shoulder) appear to be more strongly associated with prosthetic skill retention than users' cognitive status. This may be due to non-cognitive user demands (e.g., device weight) statistically masking the discrete influence of cognitive status on skill retention at higher configuration levels. Neuropsychological assessment warrants consideration as a valuable tool in rehabilitation settings to assist in functional device candidacy evaluations.