Some studies show that melinjo (Gnetum gnemon L.) seed extract contains various active ingredients that are beneficial to human health; even it has been commercialized as a health supplement product. Quality of seeds as raw material becomes one of key factors that determine the quality of product derived from melinjo seed extract. Therefore sorting becomes a critical process. However the sorting of good quality and broken seeds (moldy, chalky and perforated/infected insects) is still done manually with visual observations that tend to be inaccurate and inconsistent. This study aims to develop a new method for evaluation of quality of melinjo seeds based on digital image processing. The image is taken using two lighting systems i.e. frontlight and backlight. The results show that using color features (RGB and HSV) and certain threshold values, good quality and broken seeds can be distinguished by 92.5% and 100% accuracy using frontlight and backlight image respectively. It indicates that digital image processing can be used as an alternative for quality evaluation of melinjo seed.