The analysis of surface electromyogram (EMG) signals during voluntary isometric contractions can yield important information relating to muscle fatigue. These EMG signals are typically processed to extract specific variables such as the Mean Frequency (MNF) and the Median Frequency (MDF) and studies often follow how these parameters change through time. Traditional approaches to estimate the MNF and MDF variables are based on the periodogram, but this method suffers from a high degree of variability due in part the choice of window size, window function and other inherent limitations. In this paper we propose the use of data-adaptive filterbank spectral analysis techniques, namely the Power Spectrum Capon (PSC) and the Amplitude Spectrum Capon (ASC) methods. These new methods are shown to provide significant reductions in MNF and MDF parameter variability over a wide range of data window sizes.