Several cognitive biases are thought to play a role in the development of hallucinations. Biases in source monitoring (where participants must determine whether stimuli were internal and/or self-generated, or were external and/or other-generated) have been frequently studied, and ameliorating these biases has become a therapeutic target for some research groups. However, it is likely that the robustness of the evidence base supporting a role for source monitoring biases in the development of hallucinations has been overstated. In this talk, I will discuss inconsistencies in the evidence base, will highlight some ways in which analytic flexibility may have contributed to these inconsistencies, and will present data which demonstrates that we have been measuring source monitoring biases with suboptimal levels of reliability, in at least some contexts.