Nowadays, detection of limb movements from electroencephalogram (EEG) signals is crucial issue to focus on. Many studies on the movement detection of the limbs such as arm, leg, wrist limbs exist, while there are few studies regarding finger movements. Our study aims to determine the best classified finger pairs using binary classification with highest accuracy among individual finger movements. In this study, EEG an invasive record technique was used to record brain signals. In the visual stimulated based scenario, brain signals that occurred during individual finger movements were recorded. Muscle signals were also recorded simultaneously to capture EEG epoch the duration of the finger movements in the continuous EEG signal. Feature vectors were obtained with complexity, alpha band energy and beta band energy techniques and they were classified with support vector machine (SVM). Results from the four volunteers show thumb-little finger pair was determined as the best identified finger pair. Also, the most difficult identified finger pair was detected as the middle-ring finger pair.