Intelligent product recommendation agents (RA) are used widely in e-commerce to reduce consumers’ effort and to increase the accuracy of their decisions. This study investigates how to design RAs, for desktop and handheld devices, to alleviate the negative emotions associated with the normative decision-strategy which generates accurate decisions but only with extensive effort on the part of users. Decision-strategies and preference-elicitation methods (i.e., question and answer sessions for RAs to identify the needs of individual consumers) that are employed by RAs generate different levels of effort, accuracy, and the negative emotions, while the additional cognitive effort necessitated when using limited handheld devices moderates such relationship. Provision of the RA that mitigates the negative emotions will instigate the decision-maker to choose the normative decision-strategy for emotion-laden tasks. This study extends RA literature into the area of emotions related to decision-making and into the context of mobile computing.