This study sought (1) to retest an approach to the prediction of risk of child maltreatment and (2) to test the effect of a comprehensive prenatal and pediatric health services program on the rate of maltreatment. Of 2585 women screened at their first prenatal visit, 1154 qualified for the study. Risk assignment was determined by a structured interview. High-risk women (n = 314) were assigned to receive standard (high-risk control group; n = 154) or intervention (high-risk intervention group; n = 160) services throughout the prenatal period and during the first 2 years of their infants' life. A third group (low-risk control group; n = 295) was selected among low-risk women and received standard care without intervention services. State records were searched for substantive reports of child maltreatment up to 36 months after birth. Physical abuse was found for 5.1% of the study population; neglect was substantiated for 5.9%. Prediction efforts were effective in identifying risk of physical abuse but not of neglect. Comprehensive health services did not alter the reported abuse rate for high-risk parents and was associated with an increased number of neglect reports. Intervention reduced subject attrition and appeared to serve as a bias for detection of maltreatment. Thus this long-term, prospective approach was ineffective for child abuse prevention, perhaps because of detection biases and societal changes.