Abstract： Objective To investigate the association of fruit and vegetable intake with long-term changes of serum lipid levels in middle-aged Chinese and older Chinese population. Methods The study analyzed the data collected in the 2004 and 2007-2008 cohorts of China Multicenter Collaborative Study of Cardiovascular Epidemiology. Finally, 4 495 participants from 10 groups in various regions of China, who both were followed up in the two cohorts, were included for data analysis. They were aged 41-66 years in 2004. Fruit and vegetable consumption were collected with a simple food frequency questionnaire. The percent changes (Δ%)of serum TC, TG, HDL-C and LDL-C between the two surveys were calculated. Multivariate linear regression models were used to estimate the association of fruit and vegetable intake in 2004 with percentage changes of serum lipid levels during the two surveys. Results The proportions of individuals who consumed fruits <250, 250-499 and ≥500 g/week were 24.0%, 21.8% and 54.2%, respectively. The proportions of individuals who consumed vegetables <500 and ≥500 g/day were 76.7% and 23.3% , respectively. Compared with fruit intake <250 g/week, the regression coefficients (95% CI) of ΔTC% in 250-499 and≥500 g/week group were-1.54(-2.71,-0.37)and-1.77(-2.79,-0.76).And the regression coefficients (95% CI) of ΔLDL-C% were-2.43 (-4.39,-0.48) and-2.89 (-4.59,-1.19). Compared with vegetable intake<500 g/d,the regression coefficients(95% CI)of ΔTC% and ΔLDL-C% in vegetable intake ≥500 g/d group were-1.01 (-1.95,-0.06) and-1.83 (-3.41,-0.24). However, fruit and vegetable intake had no relationship with ΔTG% and ΔHDL-C%. Conclusion The consumption of fruit and vegetable was inversely related to long-term changes of TC and LDL-C in middle-aged and older population,but no association was found for changes of TG and HDL-C.