AbstractPurposeTo analyse the current vision screening programme for children attending school with respect to identifying children with significant refractive errors.MethodsA total of 950 children from the municipality of Roskilde, Denmark, were invited to participate in a cross‐sectional study of vision screening. 447 children aged 4.5–7 years participated in the study. The children completed a vision screening and a full eye examination. The vision screening consisted of visual acuity at distance (VA), visual acuity at distance with +2.0 glasses (VA+2), visual acuity at near (VAnear) and Lang II stereotest. A LogMAR‐based picture chart was used for the vision testing. Significant refractive errors were defined as hyperopia ≥+3.5 D, myopia >1.0 D, astigmatism ≥1.25 D and anisometropia >1.0 D.ResultsUsing receiver operating characteristic (ROC) curves, we could compare the efficiency of VA, VA+2, VAnear, VA in combination with VA+2 and VA in combination with VAnear. Area under the curve (AUC) for VA, VA combined with VA+2 and VA combined with VAnear was 0.841, 0.857 and 0.857, respectively, that is all classified as ‘very good’ screening tools, whereas VA+2 and VAnear as single screening tools were classified as ‘good’ with AUC of 0.704 and 0.775, respectively. Using the optimal cut‐off limit from the ROC curves, VA above 0.05 LogMAR as cut‐off limit showed a sensitivity of 85.2% with a specificity of 74.9%. A slightly higher sensitivity could be found when combining screening tests, but with declining specificities.ConclusionVisual acuity at distance with the use of LogMAR‐based charts is the optimal screening tool of choice; only minor improvements can be obtained combining with other tests.