This paper presents an accelerometer-based device for handwritten digit and gesture trajectory recognition. The accelerometer based device consists of basically three main parts i.e triaxial accelerometer, microcontroller and RF wireless transmission module.This parts used for sensing and collecting accelerations of handwriting and gesture trajectories of hand motions.The minimised features are passed to a neural network for recognition.The main important experimental results have successfully validated the effectiveness of the trajectory recogni-tion algorithm for the proposed device.