Motor imagery electroencephalography(EEG)is widely used in the brain-computer interface systems.Multivariate empirical mode decomposition is suitable to analyze multichannel EEG signals of non-stationary and non-linear nature.Noise-assisted multivariate empirical mode decomposition(NA-MEMD)is an advanced method which can extract task-specific frequency bands from all channels in the same scale of intrinsic mode functions(IMFs).As for a set of IMFs,how to identify IMFs that contains the most significant information is still a question of interest,and conventional approaches address it by use of prior knowledge.In this paper,a novel identification method of relevant IMFs without prior information is proposed based on NA-MEMD and Jensen-Shannon distance measure.A criterion of effective factor based on Jensen-Shannon distance(JSD)is applied to select the most significant IMF scales.At each decomposition scale,three kinds of JSDs associated with the effective factor are evaluated: between IMF components from data and themselves,between IMF components from noise and themselves,and between IMF components from data and noise.The efficacy of the proposed method has been demonstrated by both computer simulations and real world EEG data from BCI Competition Ⅳ datasets.