This paper describes an efficient biometric writer identification system that can be used in a resource contained embedded environment. Writer identification (personal identification by general handwriting) is a relatively new area of handwriting research when compared to the handwriting recognition or signature verification areas. This work is aimed at exploring only small-scale handwriting samples, especially single handwritten words. A database of such samples has been collected dynamically using a digital writing tablet and a force sensitive pen. The dynamic approach adapted in this area is largely new, and only a few papers exist on the subject. Mainly dynamic features of handwriting connected with the writing process itself are considered, although some static features are also used. It is also shown how simple parameters can be used in embedded writer identification devices, and whether they can perform adequately in this type of application. A new feature selection algorithm based on likeness coefficients is proposed. The classifiers used are Minimum-Distance Classifier, Bayes Classifier, and finally their serial combination. The efficiency of this approach and its high performance together show the reasons for employing it in embedded biometric identification devices.