Sinhala characters used in the Sinhala script by over 70% of the 18 million population in Sri Lanka, have been descended from the ancient Brahmi script. The Sinhala alphabet consists of vowels and consonants and the consonants are modified using modifier symbols to give the required vocal sounds. In the process of developing an OCR for the Sinhala script, characters are initially recognised through a multi-level filtering process using the Linear Symmetry [LS] feature [1]. The recognised character is then segmented to identify the associated modifier symbol/s. Since the use of LS recognises characters prior to segmentation, the most difficult task of separating touching characters is easily solved. A method to determine the skew angle of the script is also presented. Experiments conducted so far for widely used fonts of different sizes yield encouraging results.