Using data from the 1988 ?Two-per-Thousand National Fertility Survey in Hebei Province, this study addresses the question of how, and how well, the one-child policy in China worked during its first decade, 1979-88. Even though the Chinese government developed such strong policy measures as the birth-quota system, one-child certificate incentives, and penalties to promote the policy, they were implemented unevenly over time and among the two groups of people with differential types of household registration-peasants and workers. China's one-child policy was highly effective only among women with a worker registration, who were under greater government control. The policy measures overwhelmingly overrode the effects of socioeconomic and cultural factors on the likelihood of their having a child beyond the first. However, the one-child policy was not as effective as expected for the majority of Chinese women, who were members of households registered as peasants and lived under less government control. These women continued to have more than one child. Multivariate analysis of the fertility behavior of these women reveals that son preference strongly affected the probability of having a second and third child, even more so than the level of education, the degree of urbanization, and population policy measures.