Neural oscillations in the gamma band are of increasing interest, but separating them from myogenic electrical activity has proved difficult. A novel algorithm has been developed to reduce the effect of tonic scalp and neck muscle activity on the gamma band of the EEG. This uses mathematical modelling to fit individual muscle spikes and then subtracts them from the data. The method was applied to the detection of motor associated gamma in two separate groups of eight subjects using different sampling rates. A reproducible increase in high gamma (65–85 Hz) magnitude occurred immediately after the motor action in the left central area (p = 0.02 and p = 0.0002 for the two cohorts with individually optimized algorithm parameters, compared to p = 0.03 and p = 0.16 before correction). Whilst the magnitude of this event-related gamma synchronisation was not reduced by the application of the EMG reduction algorithm, the baseline left central gamma magnitude was significantly reduced by an average of 23 % with a faster sampling rate (p < 0.05). In comparison, at left and right temporo-parietal locations the gamma amplitude was reduced by 60 and 54 % respectively (p < 0.05). The reduction of EMG contamination by fitting and subtraction of individual spikes shows promise as a method of improving the signal to noise ratio of high frequency neural oscillations in scalp EEG.