Objectives: To describe beliefs about infant sleep position among African American grandmothers and other older caregivers (senior caregivers [SCGs]) and to measure the incremental effect on the rates of prone sleep after educating a group of African American SCGs. Design: Survey of sleep practices and beliefs. Randomized, controlled trial of a teaching intervention. Settings Managed care prenatal clinic for normal risk obstetrics patients. Participants: Low-income pregnant women, self-identified as African American. An SCG is an older family member or friend to whom they would go for advice about routine child care. Pregnant women randomly assigned to the control group (educated about safe sleep practices) or the experimental group (education for both pregnant women and their SCG). Intervention: During the third trimester, 2 teaching sessions for pregnant women (both control and experimental groups) and for the experimental group's SCGs. One-on-one teaching emphasizing that infants should sleep supine. Main Outcome Measures: During the third trimester, description of rate of preference for prone sleep for infants among pregnant women and SCGs. Effects of teaching SCGs on (1) SCGs' postnatal beliefs about sleep position as a means to reduce the risk of sudden infant death syndrome, and (2) the eventual rate of prone sleeping among study infants. Results: One hundred twenty-five women were in the control group; 98 pregnant women and SCG pairs were in the experimental group. Senior caregivers were 47.1 ± 12.4 years old (mean ± SD). Most were grandmothers (either maternal, 72.5%, or paternal, 14.3%) or aunts or sisters (10.2%). No differences were noted in the prenatal rates of the prone preference (controls, 36.3%; experimental group, 35.7%; and SCGs, 34.7%). Teaching SCGs did not increase the rate of the usual prone sleep in the experimental vs the control group (13.3% vs 17.3%, X2 = 0.59, P =.44, 95% confidence interval for difference from -5.8% to + 13.8%). After the teaching and during the pregnancy, the SCGs became less worried that the infant was susceptible to sudden infant death syndrome (X2 = 16.6, P = .003) or likely to die of sudden infant death syndrome (X2 = 24.7, P < .001). Their concerns about sudden infant death syndrome were significantly less postnatally, in particular when the infant was placed supine (X2 = 19.4, P < .001). Conclusions: Most African American women and SCGs endorsed the supine sleeping position for infants. Among a group of African American infants, prenatal teaching of their grandmothers and other SCGs did not have a statistically significant incremental effect on the rates of the usual prone sleeping position. Contrary to our starting hypothesis, the SCGs of pregnant women who receive prenatal care seemed responsive to messages about supine sleeping. The Back-to-Sleep message should be delivered prenatally, but special prenatal interventions should attempt to reach women who do not receive prenatal care and SCGs who persist in their opposition to the supine sleeping position.