This paper aims to develop a data-based Semi-Quantitative Food Frequency Questionnaire (SQFFQ) covering both urban and rural areas in the Chaoshan region of Guangdong Province, China, for the investigation of relationships between food intake and lifestyle-related diseases among middle-aged Chinese. We recruited 417 subjects from the general population and performed an assessment of the diet, using a 3-day weighed dietary record survey. We employed contribution analysis (CA) and multiple regression analysis (MRA) to select food items covering up to a 90% contribution and a 0.90 R-2, respectively. The total number of food items consumed was 523 (443 in the urban and 417 in the rural population) and the intake of 29 nutrients was calculated according to the actual consumption by foods/recipes. The CA selected 233, 194 and 183 foods/recipes for the combined, the urban and the rural areas, respectively, and then 196, 157 and 160 were chosen by the MRA. Finally, 125 foods/recipes were selected for the final questionnaire. The frequencies were classified into eight categories and standard portion sizes were also calculated. For adoption of the area-specific SQFFQ, Validity and reproducibility tests are now planned to determine how the combined SQFFQ performs in actual assessment of disease risk and benefit.