There are different feature extraction methods in brain-computer interfaces (BCI) based on Steady-State Visually Evoked Potentials (SSVEP) systems. This paper presents a comparison of five methods for stimulation frequency detection in SSVEP-based BCI systems. The techniques are based on Power Spectrum Density Analysis (PSDA), Fast Fourier Transform (FFT), Hilbert-Huang Transform (HHT), Cross Correlation and Canonical Correlation Analysis (CCA). The results demonstrate that the CCA and FFT can be successfully applied for stimulus frequency detection by considering the highest accuracy and minimum consuming time.