The accurate estimation of inflation persistence across different cities has important practical implications on making macro economic policy.This paper employs the recently developed Grid Bootstrap method to investigate the nature of inflation persistence in 36 cities in China and obtain median unbiased estimates and confidence interval with correct coverage.The empirical results suggest that the conventional estimator produces significantly downward biased estimates for inflation persistence and the largest bias is above 10%.The level of inflation persistence across different cities also manifests considerable differences.The important implications of the empirical finding are also discussed.