The exact pathogenesis of visual hallucinations in Parkinson's disease is not known but an integrated model has been proposed that includes impaired visual input and central visual processing, impaired brainstem regulation of sleep-wake cycle with fluctuating vigilance, intrusion of rapid eye movement dream imagery into wakefulness and emergence of internally generated imagery, cognitive dysfunction and influence of dopaminergic drugs. In a clinical study, we assessed motor and non-motor function, including sleep, mood, autonomic and global, frontal and visuoperceptive cognitive function in patients with and without visual hallucinations. A subgroup of patients underwent detailed ophthalmological assessment. In a separate pathological study, histological specimens were obtained from cases of pathologically proven Parkinson's disease and a retrospective case notes review was made for reporting of persistent formed visual hallucinations. An assessment of Lewy body and Lewy neurite pathology was carried out in five cortical regions as recommended by diagnostic criteria for dementia with Lewy Bodies and in brainstem nuclei. Ninety-four patients (mean age 67.5 ± 9.5 years) participated in the clinical study of whom 32% experienced visual hallucinations. When corrected for multiple comparisons, patients with visual hallucinations had significantly greater disease duration, treatment duration, motor severity and complications, sleep disturbances, in particular excessive daytime somnolence and rapid eye movement sleep behavioural disorder, disorders of mood, autonomic dysfunction and global, frontal and visuoperceptive cognitive dysfunction. Of the 94 patients, 50 (53%) underwent ophthalmological assessment. There were no differences in ocular pathology between the visual hallucination and non-visual hallucination groups. In a logistic regression model the four independent determinants of visual hallucinations were rapid eye movement sleep behavioural disorder (P = 0.026), autonomic function (P = 0.004), frontal cognitive function (P = 0.020) and a test of visuoperceptive function (object decision; P = 0.031). In a separate study, post-mortem analysis was performed in 91 subjects (mean age at death 75.5 ± 8.0 years) and persistent visual hallucinations were documented in 63%. Patients in the visual hallucinations group had similar disease duration but had significantly higher Lewy body densities in the middle frontal (P = 0.002) and middle temporal gyri (P = 0.033) and transentorhinal (P = 0.005) and anterior cingulate (P = 0.020) cortices but not parietal cortex (P = 0.22). Using a comprehensive assessment of the clinical, demographic and ophthalmological correlates of visual hallucinations in Parkinson's disease, the combined data support the hypothesized model of impaired visual processing, sleep-wake dysregulation and brainstem dysfunction, and cognitive, particularly frontal, impairment all independently contributing to the pathogenesis of visual hallucinations in Parkinson's disease. These clinical data are supported by the pathological study, in which higher overall cortical Lewy body counts, and in particular areas implicated in visuoperception and executive function, were associated with visual hallucinations.