This article presents a new eyelid localization algorithm based on a parabolic curve fitting. To deal with eyelashes, low contrast or false detection due to iris texture, we propose a two steps algorithm. First, possible edge candidates are selected by applying edge detection on a restricted area inside the iris. Then, a gradient maximization is applied along every parabola, on a larger area, to refine parameters and select the best one. Experiments have been conducted on a database of 151 iris that have been manually segmented. The performance evaluation is carried out by comparing the segmented images obtained by the proposed method with the manual segmentation. The results are satisfactory in more than 90% of the cases.