A systematic review of trials about screening older people for visual impairment found no evidence that screening improved vision. We undertook a new trial nested within a larger cluster randomised trial of multidimensional screening for people aged 75 years and over. 106 general practices were randomised to: targeted screening in which only a small proportion of participants with a range of health problems were offered visual acuity screening, and universal screening in which all participants were offered visual acuity screening. People identified with impaired vision were referred to the eye services. Around 220 participants were randomly sampled from ten practices in each group and visual outcomes measured at three to five years. The response rate to the baseline assessments was 76.1 Over one third of eligible participants died before having an outcome assessment. Of those alive, 67.8% in the targeted screening group and 57.9% in the universal group completed an outcome assessment. At outcome 37.0% (307/829) in the universal group had visual acuity of less than 6/18 in either eye compared with 34.7% (339/978) in the targeted group (odds ratio 1.11,95% confidence interval 0.76 to 1.62, P=0.58). The 25 item National Eye Institute Visual Function Questionnaire composite score was 86.03 in the universal group and 85.62 in the targeted group (difference 0.41,95% confidence interval - 1.68 to 2.50, P=0.69). Although visual impairment was common, few people benefited from subsequent intervention. Possible explanations for the lack of effect include: chance; under-detection of uncorrected refractive error and that only around half the recommendations for referral to an ophthalmologist resulted in referral.