Objective： To establish a sensitive and rapid one-step reverse transcriptase droplet digital polymerase chain reaction （RT-ddPCR） assay for detection human astrovirus （HAstV） in different food matrices. Methods： The RT-ddPCR quantitative detection assay for HAstV was established and optimized by applying the selected specific primers and probes. The specificity of this quantitative method was validated via using other foodborne viruses RNA as targets. A HAstV RNA reference material was series diluted to evaluate the quantitative detection limit， accuracy and variability of the developed assay. In order to test the detection limit and recovery rate in food， artificial positive samples were prepared by adding MS2 bacteriophage with known titer to different food matrices. Virus elution and purification method in food from CEN ISO/TS 15216-2：2013 was adopted and the virus detection amounts of two RNA extraction methods were compared. Results： The HAstV RT-ddPCR assay had better accuracy， sensitivity and smaller variability with a detection range from 105 to 100（copies/μL） and a lower limit of quantitation of 5 copies/μL. The detection limits in digestive glands of bivalve molluscan shellfish， food with hard surface， raw vegetables and soft fruits were 7.2×106-3 600 copies， 7.2×106-3 600 copies， 7.2×106-7 200 copies and 7.2×107-7 200 copies， respectively. In higher contamination level， detection amount of virus was significantly lower （P＜0.05） when RNA was extracted and purified by trizol combined with magnetic beads. A statistically significant difference （P＜0.05） in virus recovery rate among different food matrices was indicated by one-way ANOVA and Turkey multiple comparison analysis. Conclusion： The established RT-ddPCR assay in food could effectively detect HAstV in contaminated food. However， the single RT-ddPCR quantitative result could not reflect actual virus load in food which could be calculated by applying recovery rate.