One of the tasks of a forensic document examiner is to analyse a questioned handwritten document in order to determine its author. The purpose of such analysis is to determine whether the document is genuine, forged, or disguised. Document analysis plays an important investigative and forensic role in many types of crime and is used as evidence of authorship or non-authorship in a court of law. The methods used by forensic experts in document analysis are intuitively reasonable and have been derived from experience. The experts look into characteristics of handwriting (features) that are consistent in a person's normal writing. These characteristics allow the experts to compare different handwritten samples and draw a conclusion about the authorship of the questioned sample. However, the methods used are based on experience rather then scientific facts. There is a lack of scientific basis to support their methods of analysis. The aim of this thesis is to answer the question whether the methods used by forensic document examiners for handwriting analysis allow us to distinguish different writers with reasonable accuracy when the documents are written in normal (not deliberately changed) writing. In this work a number of features commonly used by forensic document examiners were automatically extracted from greyscale images of samples of characters "d", "y", "f", "t", and grapheme "th". The discriminative power of the features was assessed. The feature extraction was performed in several iterations; in each iteration new features were added, some of the previous features were removed, and extraction algorithms were improved. Automatic feature extraction accuracy of 85% was achieved in the initial experiments and later was improved to 94% by using a novel method of handwriting stroke approximation (character skeletonisation). A new method of skeletonisation for approximation of handwriting strokes on character images was developed. The method was designed to extract the skeleton which 7 ATTENTION: The Singapore Copyright Act applies to the use of this document. Nanyang Technological University Library Some interactive tools were developed as a by-product of this study to facilitate visual comparison of handwriting samples and aid in development of the skeletonisation method. Study of document examiner features in forged and disguised handwriting is a necessary next step to provide a scientific support to forensic analysis of handwriting, and is suggested as a possible direction for further research.