Background: Controversy exists about the role of visual parameters and vision in learning to read. This study aims to determine whether ocular parameters or performance on a dynamic test of visual function differs for children of differing reading ability. Methods: Two hundred and eighty-four children (mean age 9.9 +/- 1.8 years) received a vision screening emphasising binocular anomalies associated with discomfort at near (distance and near visual acuity, distance vision challenged with binocular +1 D lenses, near heterophoria, near point of convergence, stereopsis and accommodative facility). Non-verbal mentation age and reading accuracy were assessed. One hundred and six children performed a computerised task of motion coherence detection. Children were classified as normal readers (n = 195), children with dyslexia (n = 49) or learning disabled children (n = 40) based on their mentation age and their reading age. Results: There were no statistically significant differences or correlations between visual parameters and reading performance. Over thirty per cent of the children had accommodative facilities below or equal to six cycles per minute. Children with learning disabilities performed worst on the motion coherence task but this was statistically significant only when compared to the performance of dyslexics. Discussion: The lack of association between ophthalmic parameters and poor reading ability supports the view of the Committee on Children with Disabilities. However, 39 per cent of the children might be expected to experience difficulty 'reading to learn', as suggested by the American Academy of Optometry, as they showed anomalies associated with visual discomfort with prolonged reading. The motion coherence test did not differentiate dyslexics from normal readers and was worst in children with learning disability. Accommodative facility testing remained the most useful predictor of potential visual discomfort.