OBJECTIVES: We systematically investigate random utility maximization and random regret minimization modeling approaches to establish the impact of differently framed opt-out alternatives in discrete choice experiments. We hypothesize that within the same experiment, when opt out alternatives are framed as a rejection of all the available alternatives, it is likely to have a detrimental impact on the performance of RRM model, while the performance of RUM model suffers more when the opt out is framed as a respondent being indifferent between the alternatives on offer. METHODS: We used two waves of data from a discrete choice experiment (N = 227; N= 344); the first wave included an opt-out option implying a rejection of choice alternatives (i.e. none of these) while the second wave included an opt-out option implying a position of ‘indifference’ between the choice alternatives. We compared RUM and RRM models of different sophistications (e.g.; multinomial logit and mixed logit) in terms of parameter estimates, log likelihood and the Ben-Akiva and Swait test for non-nested models. RESULTS: In line with hypotheses, RUM models performed significantly better (P<0.01) than RRM models when opt–out alternative implied rejection of choice alternatives i.e. none of these. The RRM models performed significantly better (p<0.01) than the RUM models when the opt-out alternative implied a position of ‘indifference’. RRM models had difficulty in handling the ‘none of these’ opt-out alternative while the RUM models had difficulty in handling the ‘indifference’ opt out alternative as was evident by the suspiciously large value of opt-out constant in the model parameter estimates for these cases. CONCLUSIONS: The framing of opt out alternatives influences the type of behavioral framework to be considered for modeling.