Performing phantom movements with visual virtual feedback, or mirror therapy, is a promising treatment avenue to alleviate phantom limb pain. However the effectiveness of this approach appears to vary from one patient to another. To assess the individual response to training with visual virtual feedback and to explore factors influencing the response to that approach. Eight male participants with phantom limb pain (PLP) resulting from either a traumatic upper limb amputation or a brachial plexus avulsion participated in this single case multiple baseline study. Training was performed 2 times per week for 8 weeks where a virtual image of a missing limb performing different movements was presented and the participant was asked to follow the movements with his phantom limb. Patients reported an average 38% decrease in background pain on a visual analog scale (VAS), with 5 patients out of 8 reporting a reduction greater than 30%. This decrease in pain was maintained at 4 weeks postintervention in 4 of the 5 participants. No significant relationship was found between the long-term pain relief and the duration of the deafferentation or with the immediate pain relief during exposure to the feedback. These results support the use of training with virtual feedback to alleviate phantom limb pain. Our observations suggest that between-participant differences in the effectiveness of the treatment might be related more to a difference in the susceptibility to the virtual visual feedback, than to factors related to the lesion, such as the duration of the deafferentation.