This research to build a pattern recognition hijaiyah letters handwriting application. In this offline pattern recognition research, a digital handwriting input is captured by a scanner. In this research we try to recognize digital handwriting based on feature classification using Backpropagation Neural Network and Fuzzy Feature Extraction method for feature extraction. This application is built using Matlab language programming and designed using Data Flow Diagram (DFD). Meanwhile the system development life cycle method that used is Waterfall. Experiments were performed on a single Hijaiyah letters isolated from 900 samples gave results of generalization ability of artificial neural networks for 96.11% and the ability to generalize the ability of Neural Network memorization for 98.33%.