Has survey nonresponse caused scholars to overestimate political trust in China? We analyse item nonresponse for sensitive questions on trust in government from our original survey of villagers conducted in China in 2012. We also analyse nonresponse in four other comparable surveys conducted in China between 1993 and 2014. We examine the association between nonresponse to politically sensitive questions and individual characteristics such as sex, level of education, Party membership and cosmopolitanism. We find that less privileged groups may be underrepresented in survey data generally. We find mixed results regarding the association between cosmopolitanism and nonresponse. We conclude that our understanding of political trust in China has been compromised by high rates of item nonresponse, leading to artificially high estimates of trust in the centre and exaggerated accounts of the gap between trust in central and local leaders.