Current online character recognition accuracy is not perfect, and there have been many attempts to perfect the technique over the years to ensure that Tablet PCs, PDAs, and other devices that utilize text recognition are usable and ideally transparent to the user. Many systems rely on context-based recognition, as letter identification without context clues has not been proven to be as accurate. A system was devised using directionâbased stroke analysis to attempt to improve letter recognition rates. This system was able to correctly recognize the input character for 62.12% of the samples without the use of context clues. The success rate indicates that the system is not yet accurate enough to stand alone, though it could potentially be improved upon and implemented alongside context-based methods.