Phantom limb pain (PLP) is a distressing and persistent sensation that occurs after the amputation of a limb. While medication-based treatments have limitations and adverse effects, neurostimulation is a promising alternative approach whose mechanism of action needs research, including electroencephalographic (EEG) recordings for the assessment of cortical manifestation of PLP relieving effects. Here we collected and analyzed high-density EEG data in three patients (P01, P02, and P03). Peripheral nerve stimulation (PNS) suppressed PLP in P01, but was ineffective in P02. By contrast, transcutaneous electrical nerve stimulation (TENS) was effective in P02. In P03, spinal cord stimulation (SCS) was used to suppress PLP. Changes in EEG oscillatory components we analyzed using spatio-spectral decomposition (SSD) and Petrosian fractal dimension (FD). With these methods, changes EEG spatio-spectral components were found in the theta, alpha, and beta bands in all patients, with these effects being specific to each individual. These changes in EEG patterns were found for both the periods when PLP level was stationary and the periods when PLP was gradually changing after neurostimulation was turned on or off. One consistent effect was an increase in EEG fractal dimension on the side contralateral to the amputation when PLP was present, while effective stimulation resulted in a decrease in fractal dimension. Overall, our findings align with the proposed roles of brain rhythms in thalamocortical dysrhythmia, which is has been linked to neuropathic pain. The individual differences in the observed effects could be related to the specifics of treatment in each patient and the unique spectral characteristics in each of them. These findings pave way to the closed-loop systems for PLP management where neurostimulation parameters are adjusted based on EEG-derived markers.