Objective  To assess the value of the hand-held autorefractor SureSight in screening refractive errors in children 3 to 6 years old. Methods  In a cross-sectional study， 321 children with lower visual acuity than normal participated in the SureSight and Topcon examinations， under cycloplegic and non-cycloplegic conditions respectively. Data were evaluated with Bland-Altman and ROC based on the prevalence study. Results  In the comparison， it was found that the differences in spherical diopter， cylinder diopters and spherical equivalent （SE） between non-cycloplegia with SureSight （S1） and cycloplegia with Topcorn （T2） were significant. The correlation between S1 and T2 ranged from moderate（spherical diopter： r=0.59， P<0.05； SE： r=0.54， P<0.05） to high （cylinder diopter： r=0.89， P<0.05）. The ROC curve found that myopia， hyperopia and astigmatism suspects with S1 were ≤+1.13 D， ≥+1.44 D and ≥ 0.88 D when compared to T2. The Yonden index also showed better sensitivity and specificity for astigmatism （0.679） than myopia （0.298） and hyperopia （0.270）. Conclusion  As a vision screening technique to detect refractive errors in preschool children， SureSight can help identify suspected astigmatism， but had limitations for myopia and hyperopia.