Literature suggests that the selection of potential dates online may be driven to an important extent by regret-minimization. Motivated by this suggestion, this paper estimates regret-based as well as utility-based choice models on data from a Stated Choice online date selection-experiment. Dating options were defined in terms of their drinking behavior, smoking behavior, number of children and job type. Regret-based models are found to have a significantly (though not substantially) better fit with the choice-data; this holds for the overall sample and especially for the subsample of male respondents. In line with previously formulated expectations, this paper empirically shows how the regret-based model predicts lower (higher) choice probabilities than its utilitarian counterpart when choice-options have a simultaneous poor (strong) performance on multiple attributes. In addition, also in line with expectations formulated in earlier work, we show how the regret-based model predicts higher choice probabilities than its utilitarian counterpart for choice-options that perform reasonably well on each attribute rather than having a strong performance on some attributes and a poor performance on others (so-called compromise options).