Retinal vessels tortuosity is one of the important signs of cardiovascular diseases such as diabetic retinopathy and hypertension. In this paper we present a simple and efficient algorithm to measure the grade of tortuosity in retinal images. This algorithm consists of four main steps,vessel detection, extracting vascular skeleton via thinning, detection of vessel crossovers and bifurcations and finally calculating local and global tortuosity. The last stage is based on a circular mask that is put on every skeleton point of retinal vessels. While the skeleton of vessel splits the circle in each position, the local tortuosity is considered to be the bigger to smaller area ratio. The proposed algorithm is tested over the Grisan's dataset and our local dataset that prepared by Khatam-Al-Anbia hospital. The results show the Spearman correlation coefficient of over than 85% and 95% for these two datasets, respectively.