Scholars and practitioners posit that election observers (EOs) affect local beliefs about the credibility of elections. Although these effects have important implications for democratization, they remain largely unexamined at the individual level. This article applies models of Bayesian opinion updating and motivated reasoning to illuminate the conditions under which EOs change beliefs about elections. Experimental evidence from a national survey fielded immediately following the first democratic parliamentary election in Tunisia tests the argument. Two important findings emerge. First, exposure to EOs’ positive and negative statements produces a small but significant difference in individuals’ perceptions of the election on average. Second, EOs’ negative statements cause the election’s main losing partisans—who may have had weak prior beliefs that the election was credible and were likely receptive to critical information—to believe the election was significantly less credible. These findings establish a baseline for future work on how third-party monitors shape local perceptions of political processes.