Birth masculinization in China since 1980s has drawn much attentions of scholars as well as politicians. As in many other Asian countries, there have been consistently more boys born than girls in China. Given the huge population in China, an unusually high Sex Ratio at Birth (SRB) defined as the number of baby boys per 100 baby girls born during a certain period, implies a great change in the future sex structure, and thus has long-term effects on almost every aspect of the society. The earlier studies found that education has a positive effect on SRB because people with higher level of education have better access to sex-selective technology and thus tend to have higher level of SRB than the less educated. However, now the sex-selective technology is widely available and not costly to get. So whether or not to perform sex selection on the child is based on how much boys are preferred over girls. In general, people with higher education will have open mind and not be restrained by the traditional preference for son. Therefore, education should have more discernable effects on decreasing SRB. In this thesis, we study the determination of the birth masculinity with special focus on the effects of education. First, provincial and county level from the 2010 census data in China are used to analyze the determinants for the great regional differentials of SRB. We perform correlation and regression analyses. Multiple regression models are established to explain the determinants of the overall SRB as well as the SRB for the first two orders. To indicate the education level in the region, we calculate the average numbers of years of schooling of the population in the region, for the total population and for females and males. We also calculate the relative education level of female and male. Then we include the overall education and the relative sex-specific education levels into our regression models, together with other explanatory variables suggested by the earlier findings. In view of the problem that we were using the aggregate data to make inferences about relationships at the individual level, the individual level data from China General Socio Survey in 2008 is collected. With these data, we analyze how sex composition of the children is determined by parents and households characteristics. We generate a variable called degree of son preference base on the sex composition of the children in the family, assigning higher value to families with more sons than daughters. Since the dependent variable degree of son preference is a categorical variable, we established the General Ordered Response models and perform the Generalized Ordered Logit Estimations. The education variables we use here are the wife s/mother s and the husbands /father s education, indicated as categorical variables. They enter the General Ordered Response models together with other explanatory variables. From the multiple regressions, we find that education is an important determinant for SRB differentials across regions. We find that provinces with generally higher level of education tend to have lower SRB for the second births. We find female education has strong negative effect on SRB. Both provincial data and county-level data suggest that regions where women are less deprived in education are associated with lower SRB. It works especially significant on decrease the excess male births among the second children. So improvement in female education seems able to reduce SRBs substantially. On the other hand, the male advantages in education tend to increase SRBs, which is suggested by both levels of data. Analyses with the individual level data confirm some of our findings from the aggregate data analyses. We find higher level of wife s education is associated with lower level of son preference, manifested by a lower number of boys among the children in the family. And the husband s education level is likely to reinforce son preference, at least in terms of a higher ratio of sons among the children. Based on these findings we conclude that in current China, education has started to play a role of reducing people s preference for sons. The improvement in people s education level, especially the female education level, will accelerate the pace for the sex ratio at birth to return to the normal range. Most part of the data collection and organization is performed using Microsoft Excel 2010. And the correlation and regression analyses are performed using STATA 12.0.