Betting more after losses (i.e., “loss-chasing”) is a central clinical feature of disordered gambling. According to prospect theory, increasing risk-seeking following losses could arise from a failure to “re-reference.” By contrast, successful re-referencing between successive decisions closes the mental account, and any losses are regarded as final or realized; gamblers should not chase realized losses. The present study sought to test this “realization effect” among gamblers using an ecologically valid online gambling task. We were further interested in whether the effectiveness of the loss realization varied as a function of problem gambling severity. Using online recruitment of past-year gamblers stratified on the Problem Gambling Severity Index, we tested a group without gambling problems (n = 227), a group with at-risk gambling (n = 239), and a group with gambling problems (n = 223). Over a sequence of nine bets, after the sixth bet, half of the participants underwent a simulated realization procedure that entailed cashing out from the gambling website and redepositing their remaining funds on another website. The feedback comparison group were shown their account balance after the sixth bet but did not withdraw or transfer their funds. In line with the realization effect, the group with non-problem gambling significantly reduced their bet after cashing out. The realization procedure did not significantly ameliorate loss-chasing in the groups with at-risk gambling or gambling problems. We conclude that the realization effect can be elicited in an online gambling context but that stronger interventions for realizing losses may be required for people experiencing gambling problems.