Background:
China's health warnings do not meet the FCTC Article 11 guidelines. They are small
(< 35% of the pack) and do not include pictorial images, which research has shown to have greater impact. Four rounds of warnings have been implemented between 1992-2016. This study examines warning impact over 9 years (2006-15), including the third round (April 2012), which increased font size while maintaining overall warning size and replaced the English text with Chinese on the back.

Methods:
Data are from Waves 1-5 (2006-15) of the International Tobacco Control (ITC) China Survey, a cohort survey of 800 adult smokers in each of five cities, adding five rural areas at Wave 5 (2013-15) (total N=8000). GEE logistic regression models tested changes in key indicators of warning impact over time.

Results:
Over 9 years, there has been little change in impact of the Chinese warnings. Indeed, the percentage of smokers who noticed warnings 'often' actually DECREASED from 51% (2006) to 38% (2013-15) (p< .001), and fewer than one-third of smokers reported cognitions or behaviours related to quitting because of the warnings. The April 2012 revision did not improve any of the warning impact indicators. Warning noticing/salience did not differ between cities and rural areas, but rural smokers were more likely to think 'a lot' about harms of smoking (17% vs 11%, p< .001) and about quitting (14% vs 6%, p< .001) because of the warnings.

Conclusions:
China's text-only warnings continue to be extremely weak. The minor changes in 2012 did not increase warning impact. Global evidence is clear that large pictorial warnings depicting specific harms of cigarettes would greatly enhance warning impact in China, including increasing knowledge and motivation to quit among the 300M Chinese smokers. These benefits would be especially important for the large rural population, where there are fewer other sources of health information.