Objective To evaluate the factors influencing the reproducibility and validity of food group intakes measured by a food frequency questionnaire (FFQ) for Chinese men. Methods A total of 214 adult men were randomly selected from the Shanghai Men Health Study (SMHS), an ongoing cohort study conducted in urban Shanghai. Each subject who had completed the first FFQ survey at recruitment took 24-hour dietary recall (24-HDR) surveys which were conducted once a month for 12 consecutive months and a second FFQ at the end of the study. Two FFQs were administered approximately 1.2 years apart. The food items listed in FFQ and obtained from 24-HDR were categorized into 11 food groups. The factors influencing the reproducibility of the FFQ was assessed by comparing the Pearson correlation of intake levels from the two FFQs, while the factors influencing the validity was evaluated by comparing intake levels from the second FFQ with the multiple 24-HDRs. Results A total of 195 men completed all 12 times of 24-HDR and twice FFQ surveys. The median Pearson correlation coefficients of the food groups between the two FFQs in older vs younger age group were 0.62 vs 0.50; in higher vs lower education level group were 0.58 vs 0.50; in higher vs lower income group were 0.51 vs 0.50; in higher vs lower BMI group were 0.50 vs 0.49. The median Pearson correlation coefficients of the food groups between the second FFQ and 24-HDRs in older vs younger age group were 0.61 vs 0.45; in higher vs lower education level group were 0.49 vs 0.47; in higher vs lower income group were 0.43 vs 0.57; in higher vs lower BMI group were 0.48 vs 0.51. Conclusions Age is the main factor influencing the reproducibility of the food group intakes by a FFQ used in the SMHS, while the factors influencing the validity are age and family income.