A number of paradoxes are apparent in the assessment and treatment of psychophysiological insomnia and sleep state misperception. Three of these paradoxes exist as discrepancies between polysomnographic (PSG) measures and the subjective impressions regarding sleep quality and quantity. The remaining incongruity exists largely within the objective domain. In the case of subjective-objective discrepancies, patients with insomnia: (1) frequently identify themselves as having been awake when awakened from PSG defined sleep; (2) tend to overestimate sleep latency and underestimate total sleep time as compared with PSG measures; (3) appear to derive more benefit from pharmacotherapy that can be explained by objective gains. The remaining paradox pertains to the observation that hypnotic medications, by and large, do not normalize sleep architecture or produce a more 'sleep-like' EEG. In this paper, we review possible explanations for these various paradoxes, introduce a new perspective and suggest possible research avenues. The model introduced is based on the observation that beta and/or gamma activity (which have been found to be associated with cognitive processes) is enhanced in insomnia at or around sleep onset. We propose that this kind of high frequency EEG activity may interfere with the normal establishment of sleep onset-related mesograde amnesia. As a result, the patient with insomnia maintains a level of information and/or memory processing that blurs the phenomenological distinction between sleep and wakefulness and influences retrospective judgments about sleep initiation and duration.