This paper consists a return-to-poverty risk indicator system including indicators such as the per capita annual income of poverty-stricken families, the implementation of policies, and the gender of the head of the household. Using a combination of analytic hierarchy process and fuzzy evaluation method, 50 poor households with registered cards in Guangdong Province in 2020 were taken as the research objects, and quantitative analysis of the index system was carried out. The results show that the per capita annual household income is the most influential indicator, and policy coverage, policy effectiveness, and per capita annual household expenditure also have a significant impact.