Objective To establish a screening model for preschool children with vitamin A deficiency (VAD) based on nutrition epidemiological survey, purposely screen preschool children with VAD and marginal VAD (MVAD), and provide efficient screening tool for further diet intervention of VAD. Methods A random cluster sampling method was adopted and 3 kindergartens in Ba'nan District of Chongqing were randomly selected. The survey covered dietary survey, physical examination, medical history of last 2 weeks, physical measurement, and laboratory test. The screening model of healthy nutrition score based on food intake for preschoolers (HNSP) was established according to the Healthy Nutrition Score for Kids and Youth (HuSKY) method. Results The epidemiological survey showed that a total of 492 children met the inclusion criteria, including 229 boys and 263 girls. Their mean age, height, and body weight were (4.54±0.87) years, (107.50±7.20) cm, and (18.42±3.41) kg, respectively. HNSP consisted of 10 items, i.e. scores of grains and potatoes, vegetables, fruit, livestock and poultry meat, fish and other animal seafood, eggs, beans and bean products, milk and dairy, and other foods and fortified foods or supplements. The results of effectiveness analysis showed that the accumulative score of HNSP positively correlated with dietary energy and most dietary nutrients and did not correlate with anthropometric indexes (height, weight, body mass index, weight-for-age Z score, height-for-age Z score, and BMI-for-age Z score) and other blood biochemical indexes (Hb, MCHC, serum Ca, serum Fe, and serum Zn) (P>0.05), except significant correlation with plasma retinol (P<0.01). The results of accuracy analysis indicated that the accumulative score of 70 was the cut-off value for screening VAD or MVAD. The differences of screening MVAD/VAD by HNSP evaluation and gold standard method were not statistically significant (P>0.05).The sensitivity and specificity of screening were 87.98% and 17.95%, respectively. The results of reliability analysis showed that internal consistency of HNSP was good (Cronbach α coefficient=0.86). The results of repeated consistency analysis indicated that the difference of accumulative scores of two establishments of HNSP was not statistically significant (P>0.05). The repeatability of two evaluations was good (Kappa value=0.61). Conclusion The HNSP is effective and reliable for evaluating the overall dietary quality with good repeatability and consistency and is an ideal diet screening tool for preschool children. It is recommended to conduct nutrition surveys at different time and places for different people in order to revise and improve the screening method.