This paper proposes a fully-automated retinal vascular tortuosity identification based on an early hypothesis from previous researches that better Diabetic Retinopathy (DR) detection accuracy can be obtained using retinal vascular tortuosity measurement than the current DR detection algorithms. In this work, a combination of local and global tortuosity measurements using non-linear curvature-based method is computed before a predicted tortuosity condition is determined by comparing the obtained average tortuosity value with a computed threshold value, T. The retinal vascular network is classified as normal if the average tortuosity value is less than the T and vice versa. This algorithm has been tested using twenty ground truth images from the online databases; DRIVE and STARE digital fundus datasets. The results show a vague correlation of retinal vascular tortuosity with diabetic retinopathy (DR) disease. Therefore, it can be concluded that the association between retinal vascular tortuosity and DR is still in its infancy and an objective and reliable tortuosity index is clearly needed to support the earlier hypothesis. Hence, in future work, we will develop an algorithm to compute the tortuosity of retinal arteries and veins and determine their correlation with DR.