Non-stationarity is often found in session-to-session transfers of Brain Computer Interfaces (BCIs). To cope with the problem, a framework based on Common Spatial Patterns (CSP), Linear Discriminant Analysis (LDA), and covariate shift adaptation methods is proposed. Covariate shift adaptation is an effective method which can adapt to the testing sessions without the need for labeling the testing session data. This framework has been applied on one electrocorticogram (ECoG) dataset and one Electroencephalogram (EEG) dataset from BCI Competition III. Despite the different characteristics of ECoG and EEG, non-stationarity appeared in both datasets. Results showed that the proposed framework compares favorably with those methods used in the BCI Competition, revealing the effectiveness of covariate shift adaptation in tackling the non-stationarity in Brain Computer Interfaces.