Despite the powerful legal effects of restraining orders, good U.S. national statistics on restraining orders have not been publicly available. The datasets below document U.S. restraining order statistics from public sources and make national estimates of restraining orders issued and orders in effect through both civil and criminal procedures. In the U.S. across the year 2008, courts issued an estimated 1.7 million domestic-violence restraining orders. On any given day in 2008, an estimated 1.2 million domestic-violence restraining orders were active.

Terminology for restraining orders varies across jurisdictions. Restraining orders are also commonly called orders of protection, protective orders, and protection from abuse orders. These orders typically arise from allegations of domestic violence or harassment. A restraining order restrains the subject party’s communication and association with particular persons. A restraining order can deprive a person of communication with her children and the right to live in her own home. While the term “order of protection” has become much more common than the term “restraining order,” the former term is periphrastic. Restraining more directly describes the legal act in the relevant order.

Restraining orders issue from both civil and criminal proceedings. Both types of proceedings generate initial restraining orders. Initial restraining orders typically have durations ranging from one to fifteen days. Initial restraining orders often lead to final restraining orders. The final restraining order in a given proceeding has a duration of a year or longer. Final orders are sometimes called permanent orders, but they typically don’t last permanently. In a given proceeding, initial orders can be renewed before arriving at a judgment on a final order.

Orders issued differs from orders in effect. Orders issues is a flow concept measured with respect to a specific time period, e.g. the number of orders issued in a year. Orders in effect is a stock concept relevant to any point in time, e.g. average number of orders in effect on any given day during the year.

Restraining orders statistics are complex. Recognizing initial and final restraining orders, considering civil proceedings and criminal proceedings, and distinguishing between order flows and order stocks are essential for correctly interpreting restraining order statistics.

The duration of initial restraining orders depends on both statutory specifications and the extent of renewal. Initial restraining orders issued ex parte in response to civil petitions are typically specified by statute to endure no longer than from five days to three weeks.^ The mean length of statutory time limits for initial civil restraining orders is 17 days.^ Some state statutes specify that initial restraining orders remain in effect an indefinite amount of time. That approach follows that of the Model Code on Domestic and Family Violence (1994). The Model Code stated that ex parte civil orders of protection are to be effective “until further order of the court.”^ In practice, initial restraining orders, both civil and criminal, are renewed if a hearing is postponed or a case is continued. The effective duration of initial restraining orders thus varies considerably.

Some national estimates of restraining orders have been published based on data from the national restraining order registry, but these estimates are not well documented. The Federal Bureau of Investigation (FBI) established a Protection Order File at its National Crime Information Center in May, 1997. In 2002, a bulletin from the U.S. Office for Victims of Crime stated:

In fall 1998, the national registry contained only 97,136 entries, which is estimated to be less than 5 percent of the 2 million orders believed to qualify for entry.^

The bulletin did not provide any documentation or source for “2 million orders believed to qualify for entry.” As of September, 2005, the FBI’s Protection Order File contained “more than 940,000 protection order records (in 46 states, the District of Columbia and the Virgin Islands).”^ A study published in 2005 reported:

According to the FBI, between 600,000 and 700,000 records of permanent orders are entered annually into the NCIC registry. (An additional half-million records of temporary {initial} orders are also entered annually, but many of these probably involve the same parties who are the subject of subsequently entered permanent orders. Furthermore, temporary orders of protection are not represented to the degree they are issued in state courts, where they far outnumber permanent orders of protection.) The coverage of the NCIC Registry is not comprehensive: eight states do not participate at all, and many other states have incomplete coverage. In Texas, 25 percent of the counties do not report.^

The NCIC counts of protective orders provide lower bounds for national restraining order statistics. A review of these and other restraining order statistics has generated an estimate that two to three million initial restraining orders are issued per year.^ Underscoring the lack of good, public information about restraining orders, that estimate appears to be too high by a factor of roughly two.

Study of California restraining order statistics indicates significant under-reporting of restraining orders to the state restraining order registry. In California, state law requires criminal courts to issue restraining orders whenever a domestic violence offender is sentenced to probation. California law also requires that the restraining order be entered into the statewide Domestic Violence Restraining Order System (DVROS). Nonetheless, in 2003 six large counties had in DVROS a total of less than 50 restraining orders. The data indicated that many counties were not entering restraining orders into the statewide registry.^ Under-entry is likely to be even more of a problem for a national database.

A highly cited, major government-funded study generated an estimate of restraining orders issued nationally. This study, the National Violence Against Women Survey (NVAWS), showed a total of 1,131,999 “temporary restraining orders” as “average annual justice system utilization estimates for adult victims of intimate partner rape, physical assault, and stalking.”^ The relevant question from NVAWS asked about restraining orders, not initial restraining orders:

J70 Did you get a restraining order against him/her as a result of this incident?^

The NVAWS study constructed the national estimate of restraining orders by multiplying the percentage that reported “yes” to this question by an estimate of total victimizations. The estimates for total victimizations was calculated from national estimates of the number of persons victimized in the past twelve months times estimates of the annual average number of vctimizations per victim. For intimate partner physical assaults, the latter figures for men and women are 3.5 and 3.4 physical assaults per year, respectively.^ The national estimate of (temporary) restraining orders is thus based on the assumption that restraining order use is independent of multiple victimizations per year. That would be true if initial restraining orders were completely ineffective, if initial restraining orders never led to (long duration) final restraining orders, and if initial orders were issued without respect to previous use of initial orders. Those are implausible assumptions. The rationale for both initial and final restraining orders is to protect persons from further victimization. More generally, NVAWS has serious methodological problems. The national estimate of restraining orders from NVAWS, although of the right order of magnitude, isn’t actually constructed credibly. Nonetheless, it has been cited, directly and indirectly, in authoritative fora.^

The National Center for State Courts (NCSC) has made an extensive effort to gather state statistics on civil restraining orders (protection orders) in cases of domestic violence. Reporting quality varies greatly across states. From 2003 to 2010, thirteen states did not report restraining order statistics to NCSC. Those non-reporting states include California and Texas. Those two states together contain nearly 20% of the total U.S. population. In most states, initial domestic-violence restraining orders are issued ex parte with merely pro forma review. In New York State and other states, civil petitions for restraining orders are not counted at the petitioning stage as incoming civil cases.^ A hearing for a final order is procedurally established in all such cases. But a large share of such hearings are dismissed at the request of the petitioning party or because the petitioning party does not appear. Comparing NCSC statistics to state-published statistics suggests that a variety of subtle counting issues affect the counts. Nonetheless, combining NCSC statistics with state-published statistics gives a reasonably good national estimate for civil petitions for a domestic-violence restraining orders in 2008. Those data indicate that 1.2 million domestic-violence restraining orders issued through civil proceedings in the U.S. in 2008.

Statistics on restraining orders arising from criminal proceedings for domestic violence are of worse quality than statistics on restraining orders arising from civil proceedings. Statistics on criminal domestic-violence restraining orders are not generally available in state-specific court statistical publications. A national estimate for those orders, however, can be constructed indirectly. About a million persons were arrested for domestic violence in the U.S. in 2008. Other fragmentary data suggests that a third of domestic-violence arrests generate a criminal domestic-violence restraining order. Those two figures imply that in the U.S. in 2008, about 350,000 initial domestic-violence restraining orders issued through criminal proceedings.

Recognizing two procedural regularities helps interpretation and analysis of restraining order statistics. First, about 90% of initial civil petitions for domestic-violence restraining orders are typically granted. The same is true for police requests for emergency domestic-violence restraining orders. See restraining order data for Arizona, Connecticut, Florida, and Pennsylvania , as well as detailed studies of courts in Gardner, Massachusetts, Pittsburgh, Pennsylvania, and Honolulu, Hawaii. Second, only roughly one-third to two-thirds of initial restraining orders lead to a final restraining order. See restraining order data for Connecticut, Florida, Kentucky, Ohio, and Pennsylvania, as well as detailed studies of courts in Gardner, Massachusetts, Howell County, Missouri, selected Kentucky counties, Pittsburgh, Pennsylvania, Honolulu, Hawaii, Washington, DC, and for four major U.S. cities. Many scheduled hearings for a final order are dismissed with the consent of the parties or because the petitioning party does not appear. Hence the share of initial orders for which a final order is not granted is much larger than the share of initial orders for a which a judge formally rules not to grant a final order.

The national restraining order dataset includes a simple accounting model that encompasses stocks and flows of initial and final restraining orders arising from civil and criminal proceedings. In a steady-state equilibrium, flows and stocks of orders are related through average order duration. Rough estimates of average order durations for initial and final restraining orders arising from civil and criminal proceedings connects data on restraining orders issued to the the stock of restraining orders active. Because the durations of initial and final orders differ greatly, their respective shares differ greatly across orders issued and orders active.

A significant share of restraining orders in the U.S. address circumstances other than domestic violence. Domestic violence is not statutorily defined in a conceptually meaningful way. Court statistics that separate domestic violence restraining orders from other restraining orders, e.g. “injunctions against harassment,” show that other restraining orders amount to about 33% of total restraining orders (see for example, restraining order data for Arizona, Colorado, Florida, Nevada, and Washington). Other restraining orders can easily include circumstances falling formally within domestic-violence law. For example, Justice Courts in Las Vegas, Nevada, formally lack jurisdiction under law to issue domestic violence protection orders (restraining orders). However, a thorough study of protection orders that those courts issued in 2008 indicates that 30% to 50% of the Justice Courts’ non-domestic violence protection orders addressed parties with domestic relations.^ The specific share depends on how domestic relation is defined (see Nevada restraining order data for calculation details). The statistics in the datasets below include all types of restraining orders. In forming national estimates of domestic violence restraining orders, an estimate of other restraining orders was made based on the best available data. Inconsistencies and misunderstandings in defining domestic violence create the largest margin of uncertainty in national estimates of domestic violence restraining orders.

Restraining orders seem to be used much less frequently outside the U.S. than within the U.S. The Swedish justice system issued 3500 Contact Prohibition Orders in 2002.^ That’s 0.5 restraining orders per 1000 adults. In the U.S. in 2008, courts issued an estimated 7.7 domestic violence restraining orders per 1000 adults. The Swedish restraining order rate is thus more than fifteen times lower than that in the U.S. In Sweden, about half of applications for restraining orders are rejected. The maximum duration of an order is one year (an order may be renewed).^ In California in 2003, 6200 restraining orders had no expiration date at all.^ The U.S. is exceptional internationally in its harshness of punishment and its extraordinarily high incarceration prevalence. U.S. use of restraining orders also appears to be exceptional, although not well-recognized as such.

The scholarly literature on restraining orders exemplifies poor-quality public discourse. U.S. government grants have fostered the development of a research literature on how courts deal with restraining orders and how to increase access to restraining orders. That literature does not recognize effects of sensational public claims about domestic violence against women, the bitter scholarly dispute about domestic violence against men, and systemic problems in the overall functioning of the U.S. justice system. The restraining order (“protective order”) literature hasn’t even compiled credible, basic facts about the extent of restraining order use.

Restraining order datasets:

• restraining orders issued and active from civil and criminal proceedings in the U.S.
• civil petitions for orders of protection against domestic violence, as collected by the National Center for State Courts for 2003 to 2010
• state-specific compilation for domestic violence restraining orders and arrests (other than in state-specific datasets below)
• restraining orders and domestic violence in Alaska
• restraining orders and domestic violence in California
• restraining orders and domestic violence in Connecticut
• abuse claimed in restraining-order petitions, Massachusetts 1997
• restraining orders and domestic violence in New York
• restraining orders and domestic violence in Tennessee