It has been shown that in certain situations losses exert a stronger effect on behavior than respective gains, and this has been commonly explained by the argument that losses are given more weight in people's decisions than respective gains. However, although much is understood about the effect of losses on cognitive processes and behavior, 2 major inconsistencies remain. First, recent empirical evidence fails to demonstrate that people avoid incentive structures that carry equivalent gains and losses. Second, findings in experience-based decision tasks indicate that following losses, increased arousal is observed simultaneously with no behavioral loss aversion. To account for these findings, we developed an attention-allocation model as a comprehensive framework for the effect of losses. According to this model losses increase on-task attention, thereby enhancing the sensitivity to the reinforcement structure. In the current article we examine whether this model can account for a broad range of empirical phenomena involving losses. We show that as predicted by the attentional model, asymmetric effects of losses on behavior emerge where gains and losses are presented separately but not concurrently. Yet, even in the absence of loss aversion, losses have distinct effects on performance, arousal, frontal cortical activation, and behavioral consistency. The attentional model of losses thus explains some of the main inconsistencies in previous studies of the effect of losses.