Objectives Charles Bonnet syndrome (CBS) is a common cause of visual hallucinations in older people. The relationship between CBS and cognitive impairment is unclear, but anecdotal reports exist of dementia emerging in patients diagnosed with CBS. This work set out to determine if there is an increased incidence of dementia, and increased severity of cognitive impairment, in people with CBS compared to controls from the same clinical setting. Method People over 65 attending low‐vision and glaucoma clinics, and a cohort of age‐matched controls, underwent a psychiatric assessment. The cohorts were followed up after one year. Results Mild cognitive impairment was present in 2/12 CBS participants and 2/10 controls. Partial insight was seen in nine CBS participants. Two participants with CBS, and no controls, developed dementia at follow‐up. No significant differences in performance on the ACE‐R were found between the groups. Both participants who developed dementia had partial insight and hallucinations of familiar figures at diagnosis of CBS, and one had mild cognitive impairment. Conclusions Reassurance that CBS is universally benign may be misplaced. Some people given this diagnosis go on to develop dementia. Cognitive testing at the point of diagnosis was unable to identify those at risk of this outcome. Partial insight, the presence of Mild Cognitive Impairment, and hallucinations of familiar figures at diagnosis of CBS may confer an increased risk of subsequent dementia diagnosis. Copyright © 2017 John Wiley & Sons, Ltd.