Purpose: Although premature birth is recognised as a cause of cerebral visual impairment (CVI), which can include cerebral visual dysfunction (CVD), the incidence and nature of CVD in prematurely born children is not known. Methods: A prospective, controlled investigation was undertaken of forty-six, mainstream primary school children, prematurely born with gestations of 24.0–34.6 weeks, and of 130 control (term-born) children. Assessments were made of IQ, ophthalmic functions, visual perception and visual attention. Structured history-taking seeking evidence of behavioural features of CVI used a question inventory. Obstetric, neonatal and paediatric medical histories were documented from case records. Results: Fifteen out of forty-six (33 %) of the prematurely born children—“cluster A”—revealed behaviours corresponding with CVD on cluster analysis of the CVI inventory. The whole prematurely born group performed worse than controls on all visual perception tests and all four visual attention tests. Children in cluster A were responsible for this effect, performing worse than controls on all visual perception and visual attention tests except visual closure, while cluster B prematurely born children performed no differently to controls. Conclusions: The prevalence of CVD in these prematurely born children is between 21–47 % (95 % CI), with a pattern similar to “dorsal stream dysfunction”. Currently available perceptual tests appear to be unable to identify the specific pattern of problems noted in this group. Many studies have provided evidence of cognitive and intellectual dysfunction in prematurely born children, and it is possible that CVD is a contributor. The CVI inventory is a potential means of identifying and characterising the condition, which can be ameliorated with simple strategies.