An anti-theft monitoring system for transmission tower based on Independent Component Analysis（ICA） is presented to examine the vibration signals hidden in the electromagnetic interference signal to prevent the high incidences of transmission tower from being stolen. It is of great importance to guarantee the security of power systems operation. The monitoring system using the ICA algorithm is to find underlying factors or components from multivariate statistical data with the help of the dimension reduction methods available to extract the non-stationary vibration signals. The anti-interference measures are adopted in the hardware of the system so as to reduce the electromagnetic interference effectively. Moreover， the vibration signals are extracted and separated successfully by use of the Negentropy based FastICA method and the Exploratory Projection Pursuit（EPP） method. The vibration pulse is also extracted efficaciously by using the adaptive threshold pulse extraction method to suppress the interference signals to a certain extent. Tests results on the spot demonstrate the system is effective in checking the vibration signals.