Risky decisions are ubiquitous in daily life and are central to human behavior, but little attention has been devoted to exploring whether risky choice can be influenced by gaze direction. In the current study, we used gaze-contingent manipulation to manipulate an individual's gaze while he/she decided between two risky options, and we examined whether risky decisions could be biased toward a randomly determined target. We found that participants' risky choices were biased toward a randomly determined target when they were manipulated to gaze longer at the target option (Study 1, N = 37; Study 3, N = 40) or at the target outcome dimension (Study 2, N = 37). We also found that both the relative time advantage and the location of the last fixation mediated the effect of the gaze-contingent manipulation on risky choice in the valid trials. However, the mediation effects of the relative time advantage and the location of the last fixation were not significant when timed-out trials were included in Studies 2 and 3, indicating that the gaze-contingent manipulation did not effectively enforce a bias toward attending to a particular stimulus through eye gaze in all trials. Future work is needed to improve the effectiveness of the gaze-contingent prompt procedure.