This paper estimates the gender wage gap and its composition in China's urban labor market using the 2009 survey data from the Chinese Family Panel Studies. Several estimation and decomposition methods have been used and compared. First, we examine the gender wage gap using ordinary least square regression method with a gender dummy variable. Then, we apply Oaxaca (1973) decomposition method with different weighting systems to analyze the logarithmic wage differential. To be more specific, we prove the existence of sample selection bias caused by the female's labor force participation. We eliminate it by using the Heckman's two-step procedure. Empirical results reveal that male workers generally receive a higher wage than female workers, and a great deal of this difference is unexplained. Meanwhile, this unexplained part, which is usually referred to as discrimination turns out to be higher when the adjustment is made for the selection bias. A further breakdown of the wage gap shows that among all the individual characteristics, occupations explain the largest share of the wage gap, followed by their working experience. On the other hand, education acts as a contributor for discrimination in the labor market.