Two studies were conducted to examine the relation between the gambler’s fallacy and attentional processes associated with inhibition of return (IOR). In Study 1, participants completed rapid aiming movements to equally probable targets presented to the left and right. They also completed a gambling protocol in which they bet on the illumination of either target. Consistent with the inhibition of return phenomenon, participants were slower to initate their movements on trial N + 1 when the target was the same as trial N. Participants with more pronounced IOR were more likely to switch betting behavior after a win than participants with a smaller index. This betting behavior was also related to a gambler’s fallacy index measured by a questionnaire. In Study 2, participants performed both the aiming task and the betting task with a partner. Each participant performed two trials before ceding to the partner. Thus we were able to examine IOR and betting behavior as a function of the participant’s own previous trial and their partner’s previous trial. The IOR effect was robust both within and between-participants. Participants were more likely to maintain their bet following an unsuccesful outcome regardless of whether it was their own outcome or their partner’s outcome. This type of betting behavior is consistent with the gambler’s fallacy. Individual IOR scores were a reliable predictor of betting behavior and the questionnaire was also succesful in predicting behavior. In addition, the within-person IOR indices covaried with the gambler’s fallacy index deprived from the questionnaire. In summary, there appears to be a relation between inhibition of return and the gambler’s fallacy. We suggest that early humans developed specialized attentional systems to deal with non-random environmental contingencies, and that the automatic processes associated with these systems are sometimes maladaptive in artifical environments in which the same contingencies do not hold.