Background: The Kish grid is commonly used in risk factor surveillance as a sampling method within households,. However through the risk factor surveillance surveys of Jiangsu Province of China in the years of 2007, 2010 and 2013, in which Kish grid method was used to select one eligible respondent within a household, we observed that the structure of samples was different from the population. Purpose: To determine if the Kish grid method distorts the structure of the sample in risk factor surveillance and to explore possible solution to this problem.