Narrative persuasion, i.e., the impact of narratives on beliefs, behaviors and attitudes, and the mechanisms underpinning endorsement of conspiracy theories have both drawn substantial attention from social scientists. Yet, to date, these two fields have evolved separately, and to our knowledge no study has empirically examined the impact of conspiracy narratives on real-world conspiracy beliefs. In a first study, we exposed a group of participants (n = 37) to an X-Files episode before asking them to fill in a questionnaire related to their narrative experience and conspiracy beliefs. A control group (n = 41) had to answer the conspiracy beliefs items before watching the episode. Based on past findings of both the aforementioned fields of research, we hypothesized that the experimental group would show greater endorsement of conspiracy beliefs, an effect expected to be mediated by identification to the episodes' characters. We furthermore hypothesized that identification would be associated with cognitive elaboration of the topics developed in the narrative. The first two hypotheses were disproved since no narrative persuasion effect was observed. In a second study, we sought to replicate these results in a larger sample (n = 166). No persuasive effect was found in the new data and a Bayesian meta-analysis of the two studies strongly supports the absence of a positive effect of exposure to narrative material on endorsement of conspiracy theories. In both studies, a significant relation between conspiracy mentality and enjoyment was observed. In the second study, this relation was fully mediated by two dimensions of perceived realism, i.e., plausibility and narrative consistency. We discuss our results, based on theoretical models of narrative persuasion and compare our studies with previous narrative persuasion studies. Implications of these results for future research are also discussed.