The two studies found that surveillance bias in the course of measuring the outcomes of child-maltreatment interventions occurred only under a limited set of conditions and could be minimized even under those conditions by identifying and incorporating data on surveillance into the analysis. Interventions that involved home visits did not detect more frequent or less severe reoffending in the course of an intervention compared to interventions that involved services to families at centers. Surveillance bias was significantly greater during periods when participants received services that increased their interactions with service providers, regardless of the content or delivery mechanisms for such services. This report suggests how to deal with surveillance bias under such conditions so as to produce an accurate measurement of the intervention's impact. Data for one of the studies were collected from three family-services outcome studies that examined child welfare reports made by members of the staff providing services to the families. The services received across participants varied and involved both prevention and intervention objectives. Data for the second study were collected from an evaluation of a statewide home-visiting prevention program for first-time mothers who were enrolled in the program during pregnancy. The home visits were conducted by nurses, who were trained to report any suspicions of child maltreatment. Follow-up child welfare reports, beginning at the point of childbirth, were submitted to the child welfare report database for both program participants and a comparison group. 5 tables, 2 figures, and 15 references