This paper presents the use of genetic algorithm in old document image enhancement and restoration. The term of 'old document' is document coming from hundreds years ago. The purpose of the study is to preserve the information contained in old documents into a digital form, because the process to rescue the old images using physical approach is too slow. The main contribution of this study is in the development of its fitness function to select new population and the best new enhanced image. The experimental results also show the influence of using the median filter in the image preprocessing. The results show relatively effective image enhancement (92.9% data get more than 90% success rate), and in general the effect of using the median filter preprocessing is not good that makes blurring effect to the images (only 59.5% data get success rate above 90%).