Background: Auditory hallucinations are frequently conceptualized as a disorder of input, whereby random discharges in language-related cortical areas lead to sensory irritations that mimic real voices. Alternatively, auditory hallucinations may represent a disorder of interpretation, whereby none of its four prevalent characteristics (the "four A's of hallucinations": acoustic, alien (i.e., appears as non-self), autonomous (i.e., beyond subjective control), authentic (i.e., appears like a real voice)) can reliably discriminate real versus imagined voices. Method: The study explored the resemblance between imagined (i.e., auditory hallucinations) and real voices. Further, the cognitive and sensory profiles of thoughts, intrusions/obsessions and voice-hearing were examined. To circumvent conservative response biases, an Internet study was conducted. 160 subjects completed the survey. Of these, 45 were diagnosed with schizophrenia, 55 had obsessive-compulsive disorder (OCD) and 60 were non-clinical controls. Results: In line with prior research, most schizophrenia patients and approximately every 7th non-clinical and every 7th OCD participant reported hearing voices. The results lend support to the claim that none of the four A's of hallucinations is specific to voice-hearing and therefore challenges the assumption that this class of phenomena reflects a false but reasonable inference of anomalous input. Importantly, a large number of voice-hearers (37%) admitted that their voices did not appear very real, and that they were less loud than real voices (52%). Voice-hearers, irrespective of diagnostic status, reported greater vividness and loudness of mental events even for normal thoughts and obsessions suggesting that enhanced mental vividness, in addition to the presence of metacognitive biases, may represent vulnerability factors for the development of hallucinations. Conclusions: Differences between intrusions and voice-hearing are more quantitative than qualitative, supporting the view that voice-hearing is more than a disorder of input. The results do not completely refute a bottom-up account of voice-hearing but suggest the involvement of important top-down attributional processes.