PIP: Described are the trends and levels of infant mortality in China by sex, time period, region, and ethnic group. Data are obtained from the 1982 census, the 1990 census, and the 1988 2% sample survey on fertility. Jiang et al. national and provincial life tables of infant mortality (IM) from 1982 census data indicate that 35.56% of male infants and 33.72% of female infants died in 1981. The 1982 estimates were considered underreports. This underreporting was confirmed in a national sample survey among 62 sampling areas conducted by Zhou Shangyou et al. (1989). Zhou et al. found that IM in urban areas was underreported by 4%, IM in rural areas was underreported by 44%, and national rates were underreported by 27%. Reporting in the 1990 census also indicated underestimates and great disparity between provinces and regions in the degree of underreporting. Three different approaches yield different estimates of infant mortality, but all estimates indicate higher female infant mortality. One approach accounted for underreporting of mortality by age, sex, time, and geographic area before setting up national and provincial life tables. This approach resulted in adjusted mortality rates of 32.19% for male infants and 36.83% for female infants in 1989. The 1988 survey produced the most reliable estimates of infant mortality. Yan and Chen (1991) produced national life tables during 1945-84 that are considered the only reliable source of national infant mortality data. This analysis relies on Chinese Population Yearbook for 1982 and China Population Newsletter for 1975-87 rates, Jiang's adjusted rates for 1990, Huang's rates by ethnic group, and Yan and Chen rates. Infant and child mortality declined rapidly over the decades. Excess female infant mortality occurred during the 1960s and 1980s and among most provinces and some ethnic groups. Sex differences by region are attributed to son preference, sex inequality, socioeconomic development, and fertility.