AbstractObjectives: Deficits in the recognition of negative emotions emerge before clinical diagnosis in Huntington’s disease (HD). To address emotion recognition deficits, which have been shown in schizophrenia to be improved by computerized training, we conducted a study of the feasibility and efficacy of computerized training of emotion recognition in HD. Methods: We randomly assigned 22 individuals with premanifest or early symptomatic HD to the training or control group. The training group used a self-guided online training program, MicroExpression Training Tool (METT), twice weekly for 4 weeks. All participants completed measures of emotion recognition at baseline and post-training time-points. Participants in the training group also completed training adherence measures. Results: Participants in the training group completed seven of the eight sessions on average. Results showed a significant group by time interaction, indicating that METT training was associated with improved accuracy in emotion recognition. Conclusions: Although sample size was small, our study demonstrates that emotion recognition remediation using the METT is feasible in terms of training adherence. The evidence also suggests METT may be effective in premanifest or early-symptomatic HD, opening up a potential new avenue for intervention. Further study with a larger sample size is needed to replicate these findings, and to characterize the durability and generalizability of these improvements, and their impact on functional outcomes in HD. (JINS, 2017, 23, 314–321)