PIP: The author argues that the effect of sex preference must be disentangled from the effect of number preference in Korea. This study tests--with hazard models--the effect of the number of previous children on the next birth according to the sex composition of previous children. Data were obtained from the 1974 Korean Fertility Survey. This paper also analyzes the timing of childbearing in recent periods in order to determine whether replacement-level fertility is temporary or permanent. The ideal number of children declined from 3.9 children in 1965 to 2.1 in 1991. The age-specific fertility rates for ages 20-24 years declined rapidly during the late 1980s. The fertility rates among women aged 25-29 years and 30-34 years increased during 1985-90. The proportion of fertility among women aged 20-29 years increased from 67.9% in 1975 to 86.6% in 1984. Women born in the late baby boom period of the late 1950s to mid-1960s reached prime reproductive age during the late 1980s and 1990s, but the crude birth rate remained about the same during 1985-92. A higher percentage of women (22.4%) born during 1955-59 remained single in 1990. During 1960-90, the percentage of women aged 20-24 years who were married declined. These trends indicate later age at childbearing and an explanation for the temporary nature of below-replacement fertility in the late 1980s. Korean women did not want to have more than 2 children, and the interval between first and second births increased since 1985. Among pregnancies of parity 2 conceived since 1985, over 90% of women with at least one son ended subsequent pregnancies by abortion compared to only 59% without sons. Hazard models of 1974 data reveal that son preference had an important effect on fertility. Fertility was higher among women with only daughters. Findings suggest that the value of sons must be measured at the societal and not at the individual level.