Exposure to violence, particularly domestic violence (DV), negatively affects children's physical, emotional, and cognitive well-being. The American Academy of Pediatrics recommends routine DV screening of female caretakers of pediatric patients. Few reports of screening in pediatric practices exist, and none have reported outcomes from a resident-run urban academic center. We set out to determine whether the use of the Partner Violence Screen (PVS) increases detection of DV and to test the mechanics of implementing large-scale DV screening in a busy, pediatric residency training clinic. Using the PVS, we screened a sample of consecutive female caretakers/guardians of children seen for pediatric care in the general pediatric clinic of Children's Hospital of Michigan from March 1, 2002, through February 28, 2003. Positive screens obtained during the study period were compared with the number of DV referrals received by the clinic social workers from January 1, 2001, through December 31, 2001, before PVS screening began. To test the mechanics of screening, we also analyzed the number of forms returned blank or marked "no opportunity to screen" in the last 8 months of the study period. In the 12 months before use of the PVS, our social work department received 9 referrals because of DV from the general pediatric clinic, among a total of 5446 caretakers/guardians bringing 6380 children for a total of 13,576 patient care visits. In contrast, the social work department received 164 referrals because of positive screening results among 5445 caretakers/guardians bringing 7429 children for 17,346 patient care visits in the 12-month study period after introduction of the PVS. Fourteen of 164 positive PVSs were found to involve nondomestic violence perpetrated by nonpartners or violence with the patient as the victim, not the mother or female caretaker. A total of 150 PVSs involved true DV. The difference in identification of DV with the PVS, compared with the rate before its introduction, was highly significant. The positive predictive value for the PVS was 91.5%, and the identified prevalence rate was 3.7%. In the last 8 months of the study period, 6301 of 8055 PVS forms (78%) were completed; 1754 of 8055 PVS forms (22%) were left blank, but it was not possible to determine whether these represented duplicate screening forms for instances in which the mother or female caretaker had brought >1 child for care. Formal screening for DV with the PVS in this study setting of a busy, urban, academic, general pediatric clinic appeared to be very successful, in terms of increasing referrals and documentation of previously unrecognized DV situations. This increase signals the need for resources (time and/or social work services) to provide appropriate referral services. The PVS identifies nonpartner violence occasionally.