Public health surveillance seeks answers to five questions: who, what, when, where, and why? For most health events, answers can be readily obtained from the patient’s medical record. Injuries are a different story. Medical records are generally adequate for telling us who was injured, what type of injury was sustained, and when the injury occurred. Unfortunately, the medical record is less useful for telling us where the injury took place, and most importantly, why.1 This is particularly true of injuries caused by firearms. There are several reasons why medical records alone are inadequate for local surveillance of firearm injuries. First, emergency physicians and trauma surgeons rarely bother to record information that is not immediately relevant to the care of their patients. The location and circumstances of an incident may be interesting to an epidemiologist or a police investigator, but they have little bearing on the decision to take a patient to the operating room. Second, many shooting victims are too injured, or too chemically impaired, to provide a reliable history at the time they are brought to the emergency department (ED). When confronted with a critically injured patient, ED personnel treat first, and worry about the history later. Finally, some victims provide inaccurate or misleading information to avoid arrest or retaliation. In one study, 25% of firearm assault victims claimed that their shooting was unprovoked.2 The difference between “victim” and “offender” is sometimes defined by who shot first, or had better aim. Without detailed information about the location and the circumstances surrounding each shooting, firearm injury surveillance provides little more than wound and body counts.3 This may suffice for national and statelevel surveillance, but it is inadequate for local problem-solving. Community officials (especially heads of public safety agencies) need more complete data to plan, implement, and evaluate counter-measures. To generate this level of detail, hospital records must be supplemented by other sources of data.2,4,5 Emergency Medical Services (EMS) records can provide useful information about patients who were transported by ambulance, but EMS records miss victims who reached the hospital by foot or private vehicle. Medical examiner records capture detailed information about victims who are found dead on the scene, but ME records are limited to fatalities. In urban areas, nonfatal shootings outnumber firearm suicides, homicides, and unintentional deaths by a ratio of four to one.2 Law enforcement records are the best source of information about what happened on the scene. Police respond to the vast majority of shootings, and they have a legal responsibility to determine what happened. They routinely interview witnesses and others who have information about the shooting, and they document their findings. As a result, police reports contain valuable information about incident location, the circumstances surrounding the shooting, victim-offender relationship, and the type of weapon involved.5 Unfortunately, it is not always easy to gain access to police files. Law enforcement agencies are notoriously reluctant to share information, especially to “outsiders.”6 In the absence of a legal obligation to share data, it may be difficult to convince them that the effort is worth the trouble. Once access is granted, logistical barriers may hamper acquisition and linkage of data. Many law enforcement agencies lack sufficient resources to computerize their records,7 and budget cuts have forced many police departments to reduce the size and scope of their crime analysis units. Manual retrieval of records is a tedious and uncertain process. Since most departments do not allow outsiders to search their files, this can greatly prolong the time required to answer requests for data. The high cost of photocopying, and the poor legibility of hand-written records pose additional problems. The task is much simpler when reports are stored in an electronic format. Officials at the California Department of Public Health obtained computerized copies of their state’s Federal Bureau of Investigation (FBI) Uniform Crime Reports (UCR) Supplemental Homicide Reports (SHR), and linked these records to state death certificates.8 Unfortunately, UCR data cannot be used to supplement documentation of nonfatal firearm injuries, because the UCR From the Center for Injury Control, Rollins School of Public Health of Emory University, Atlanta, Georgia 30322. Address Correspondence to: Arthur L. Kellermann, MD, MPH, Center for Injury Control, Rollins School of Public Health, 1518 Clifton Road, NE, Atlanta, Georgia 30322.