This paper investigates the effects of family background on academic achievement in basic education (grade 1-9) in rural China, using information on a sample of children aged 9-12 in 2000 from Gansu, China. The instrumental variable method developed by Mason and Griliches (1972), and Blackburn and Neumark (1992) is applied to control for unobserved child ability. Scores of a cognitive ability tests are first used to proxy unobservable child innate ability. This error-ridden measure of child innate ability is then instrumented by an instrumental variable generated by the Great Famine in China, 1958-1961. Empirical results indicate that omission of child innate ability leads to overestimation of income effects. Parental education is found to be key determinants of student achievement, but the roles of father’s education and mother’s education differ across child gender and levels of ability. For example, father’s education has significantly positive effect on academic achievements for both boys and girls, while mother’s education only matters for girls. The effect of father’s education matters for lower ability children, while mother’s education matters for higher ability children.