Vision is the most complex human sense and a very important instrument for communication. A lot of knowledge about a person can be obtained by studying her eyes and it is possible to trace her true intent by observing her eye movements. In the same way as speech or handwriting analysis require accurate interpretation of the speech or the pen movements, visual impression analysis require accurate interpretation of the eye movements. In this thesis the gaze tracking system Smart Eye Pro, developed by Smart Eye AB, is evaluated in terms of its use in distinguishing small eye movements. The large amount of data obtained from the gaze tracker is refined and the thesis aims at evaluating the gain of interpreting the eye movements using hidden Markov models. An eye typing application is developed to facilitate the evaluation. In this application, the user forms a word by fixating characters on a screen in front of her. Eye typing is an interesting application since the quality of the interpretation is easy to determine and the application may be an important aid for people with a handicap that impair their ability to write.