This study aims to find out the effects of mediated and interpersonal communication on voters' decision-making during a presidential election campaign. Previous election studies ordinarily assumed that the effect of communication is homogeneous throughout the election cycle. Since election campaigns are time-bound, the effects of communication may not be the same across the election campaign. This study illuminates the role of communication on voters' decision-making using event history analysis. The results show that: (1) the effects of communication are regulated or moderated by time. The effect size was observed to grow as the election season draws to a close. (2) There is a difference in mediated and interpersonal communication effects at different stages of the election campaign. Before the national convention, greater mass media exposure was found to be related to faster change in decision status from undecided to decided. However, two weeks before election day, more interpersonal communication led to a faster change in voting decisions. (3) The level of campaign interest was found to regulate the effect of communication at each campaign stage. The effect of mass media exposure grew with increasing levels of campaign interest before the national convention while that of interpersonal communication became stronger as level of campaign interest declines two weeks before election day. The results imply that the usual assumptions of election studies should be re-considered to fully understand the impact of communication on voters' decision-making. Future studies should include the effects of time and different modes of communicating in their analytical models.