This document describes the implementation of backpropagation neural networks using MATLAB. It discusses the mbackprop program, which implements the backpropagation algorithm in MATLAB. The document compares the speed of mbackprop to other backpropagation programs written in C and finds that mbackprop runs 4.5-7 times faster. It also explores how reducing the number of iterations in the algorithm can increase execution speed. Integrated graphical capabilities and other features of mbackprop are highlighted.