Control valve stiction is the most commonly found valve problem in the process industry. Quantification of the actual amount of stiction present in a loop is an important step that may help in scheduling the optimum maintenance work for the valves. In this paper, a Neural-network based stiction quantification algorithm is developed. It is shown that the performance of the proposed quantification algorithm is comparable to other method whereby accurate estimation of the stiction amount can be achieved even in the presence of random noise. Its robustness towards external oscillating disturbances is also investigated.