Table of Contents

If mistaken for an attack, a ballistic missile test, missile defense test, or space launch could spark escalation. This risk is not hypothetical. In January 1995, Russia mistook a sounding rocket launched off the Norwegian coast for a U.S. nuclear-armed ballistic missile. Because Moscow feared that the first wave of a U.S. campaign to destroy its nuclear forces might comprise a small number of ballistic missiles—perhaps just one—fired against key command-and-control nodes, the result was a cascade of warnings that reached Russian president Boris Yeltsin within four minutes.1 Only after he activated his “nuclear briefcase” and ordered the Strategic Rocket Force to prepare to launch ICBMs did the Russian military determine that the launch was benign.

Twenty-five years later, notwithstanding subsequent improvements in Russia’s early-warning capabilities, there remains a real risk that escalation could result from either a mischaracterized launch or, relatedly, from preparations for a test launch being mistaken as preparations for an attack—a scenario that would invite a preemptive strike against the launch site. There are two reasons to take these dangers seriously.

First, while the risks of escalation may be low in peacetime, they could rise significantly at times of heightened tensions. During such periods, national and military leaders might be more inclined to interpret an ambiguous event in the worst possible light because they were expecting or, at least, were concerned about an attack.2 Moreover, a state that feared that an attack might be underway or imminent would be more likely to respond precipitously in a crisis than in peacetime. Second, currently, the danger of inadvertent escalation resulting from a test or space launch (or its preparations) would be limited to a crisis involving Russia and the United States because they are the only two states with the capability to detect a ballistic missile attack and launch some of their own nuclear forces before the incoming weapons had detonated. China, however, is now acquiring a similar launch-under-attack capability, creating the possibility that these dangers could soon arise in a U.S.-Chinese crisis too.

Notifications before test or space launches can help reduce these risks. There are two operative notification agreements involving Russia and the United States: the 1988 U.S.-Soviet Ballistic Missile Launch Notification Agreement, which was made legally binding through its incorporation into New START, and the multilateral 2002 Hague Code of Conduct Against Ballistic Missile Proliferation, a politically binding document that calls for notifications of space launches (among many other provisions). Separately, in 2009, China and Russia concluded their own legally binding notification arrangement, the Agreement on Notifications of Launches of Ballistic Missiles and Space Launch Vehicles, which was extended in 2020 for ten years.3 (See table 1 for a comparison of the notification commitments.)

This patchwork of agreements has five particularly notable gaps, however. First, China and the United States have not agreed to exchange any launch notifications.4 Second, as indicated in table 1, the range thresholds that trigger notification requirements are generally quite long. Yet tests of shorter-range missiles could also spark escalation if conducted from ships or aircraft close to an adversary’s borders or from the territory of, or in the direction of, a U.S. ally. Third, no state has committed to providing notifications about tests of boost-glide missiles. Because boost-glide missiles are maneuverable, their tests are actually more likely to be misinterpreted as attacks than tests of ballistic missiles, which fly along predictable trajectories after burnout. Fourth, no state has agreed to notify others of missile defense tests, even though either the interceptor or the target missile, which could follow a ballistic or boost-glide trajectory, could be interpreted as a threatening offensive missile prior to interception (or self-destruction, should interception not occur). Fifth, there is currently no requirement on any state to provide notifications of any sub-orbital space launch—that is, a launch that places an object on a trajectory that returns to Earth, as opposed to a trajectory that takes it into Earth orbit or outer space. Yet such tests, which can be conducted for scientific research or for the development of direct-ascent anti-satellite weapons, can be misinterpreted as attacks—as the 1995 sounding rocket incident demonstrates.

Table 1: Comparison of Launch Notification Regimes
Notification Requirement China-Russia
(2009 agreement)
Russia-United States
(1988 agreement)
China-United States
Minimum range of ground-launched ballistic missile 2,000 kilometers (1,200 miles) 5,500 kilometers (3,400 miles) No notifications required
Minimum range of submarine-launched ballistic missile 2,000 kilometers 600 kilometers (370 miles)
Minimum range of air-launched ballistic missile 2,000 kilometers No notification required
Boost-glide missile test No No
Missile defense test No No
Direction of test Toward the other statea Any
Post-launch notification Yes No
Space launch to Earth orbit or outer space Yes Yesb
Sub-orbital space launch No No
Exemption for special cases Yes No

aChina is required to notify Russia of launches to the west, northwest, north, and northeast. Russia is required to notify China of launches to the northeast, east, southeast, and south.

b Pursuant to the Hague Code of Conduct.

Solution Concept

These deficiencies, paired with China’s increasing capability to detect missile launches, suggest that China, Russia, and the United States should share an interest in developing a more comprehensive approach to launch notifications.5 The proposal for a trilateral regime described below includes elements from the 1988 U.S.-Soviet and 2009 Chinese-Russian agreements, as well as from a never-implemented 2000 memorandum of understanding between Russia and the United States to enhance their notification regime. Helpfully, all these agreements contain various identical or almost identical definitions and rules (such as the instructions for describing the planned impact area of a missile test).

The following proposal aims to address all five lacunae. Notifications would be required for launches that exceed a certain minimum threshold for planned distance, apex altitude, or speed, depending on the type of launch (the risks of escalation from tests that do not meet the threshold would be small). In the case of U.S. missile defense tests, these thresholds would require Washington to report on tests of SM-3 interceptors and Ground-Based Interceptors, but not on slower systems, such as Terminal High Altitude Area Defense interceptors (of course, the thresholds would apply to all participants of the agreement equally).

A Trilateral Missile Launch Notification Regime

China, Russia, and the United States should agree to notify one another of the following:

  • All space launches
  • All test launches of ballistic or boost-glide missiles—whether conducted from air, land, or sea—that meet a specific condition:
    • For tests of ballistic missiles: The planned distance between the launch point and the impact point exceeds 500 kilometers (310 miles), or the planned apex altitude exceeds 500 kilometers.
    • For tests of boost-glide missiles: The planned distance between the launch point and impact point exceeds 500 kilometers or the planned maximum speed exceeds 2 kilometers per second (1.2 miles per second).
  • All test launches of missile defense interceptors conducted from air, land, or sea, and all launches of target missiles used in such tests, if the planned trajectory for either the interceptor or the target missile meets a specific condition:
    • For missile defense interceptors and target missiles intended to simulate the trajectory of ballistic missiles: The planned distance between the launch point and extrapolated impact point exceeds 500 kilometers, or the planned extrapolated apex altitude exceeds 500 kilometers.
    • For target missiles intended to simulate the trajectory of boost-glide missiles: The planned distance between the launch point and extrapolated impact point exceeds 500 kilometers, or the planned maximum speed exceeds 2 kilometers per second.

Both pre-launch notifications and post-launch notifications (or, if a launch did not take place, a cancellation notification) should be provided.

Pre-launch notifications should be provided at least twenty-four hours before the start of the launch window and should include the following:

  • The type of launch (space launch, ballistic missile test, boost-glide missile test, or missile defense test)
  • The total number of launch systems to be launched
  • The basing mode (ground-launched, sea-launched, or air-launched) of each launch system
  • The launch area of each launch system (for ground-based or air-based launches, the site, facility, or range; for sea-based launches, the ocean quadrant or body of water, such as a sea or bay)
  • The planned payload impact area of each launch system, if there is one; otherwise the launch azimuth (the size of the impact area may be determined by the notifying state at its discretion)
  • The time and date for the start and end of the launch window (which may last no longer than seven days, unless extended through a notification)

A single pre-launch notification may be used for multiple launches only if the last launch in the sequence is planned to occur less than sixty minutes after the first launch and if all launches are of the same type.

A post-launch notification should be provided no more than forty-eight hours after the launch and should include the following:

  • The number of launch systems that were launched
  • The date and time of the launch or launches

In implementing these provisions, the following definitions would apply:

  • “Ballistic missile” means a weapon-delivery vehicle that has a ballistic trajectory over most of its flight path and is designed to counter objects located on the Earth’s surface.
  • “Boost-glide missile” means a weapon-delivery vehicle that sustains unpowered flight through the use of aerodynamic lift over most of its flight path and is designed to counter objects located on the Earth’s surface. A reaction control system designed to change a vehicle’s attitude is not considered capable of powering flight.
  • “Extrapolated apex” means the apex of a missile defense interceptor’s or target missile’s trajectory should neither interception nor self-destruction occur.
  • “Extrapolated impact point” means a missile defense interceptor’s or target missile’s impact point should neither interception nor self-destruction occur.
  • “Missile defense interceptor” means a weapon that is designed to counter ballistic missiles or boost-glide missiles or their elements in flight.
  • “Launch system” means a space launch vehicle, ballistic missile, boost-glide missile, missile defense interceptor, or target missile.
  • “Target missile” means any vehicle launched during a missile defense test that is used as the target for an interceptor.
  • “Ocean quadrant” means a ninety-degree sector encompassing approximately one-fourth of the area of the ocean.
  • “Space launch” means a rocket launch for the purpose of delivering an object into outer space, Earth orbit, or a sub-orbital trajectory with a planned apex altitude greater than 500 kilometers.

If a launch system meets the definitions for both a missile defense interceptor and a ballistic missile or boost-glide missile, then, for notification purposes, it should be classified as a ballistic missile or boost-glide missile if its intended target is located on the Earth’s surface, and otherwise as a missile defense interceptor.


A verification system would not be needed for this proposal. In fact, this proposal is valuable precisely because Russia and the United States have sophisticated capabilities to detect missile and space launches and China is rapidly acquiring them, creating the risk of a launch being detected and misinterpreted.


Technical feasibility. This proposal should be straightforward to negotiate and implement. The one required innovation would be a mechanism for China and the United States to exchange notifications.

Political feasibility. Through their participation in existing launch notification regimes, China, Russia, and the United States have all recognized the risks of escalation as a consequence of a misinterpreted test or space launch. This proposal would benefit each state by closing a number of significant gaps in those regimes.

The political obstacles facing this proposal are relatively small—at least compared to other concepts for engaging China—since no limits would be placed on any capabilities or activities. Nonetheless, when it comes to trilateral arms control, even relatively small barriers are large in absolute terms, primarily because of the poor state of U.S.-Chinese relations.

First, both China and the United States are more concerned about deliberate aggression than they are about inadvertent escalation, increasing the difficulty of generating political traction. Yet both states have recognized the possibility that escalation may not be deliberate; for example, they both participate in the 2014 Code for Unplanned Encounters at Sea, which aims to reduce the risks associated with ships operating in proximity to one another. An even more relevant precedent was set when China announced its missile tests in advance during the 1995–1996 Taiwan Strait Crisis.6

Second, even if Beijing does not object to the principle of providing launch notifications to Washington, it may fear that doing so will lead to more pressure to engage in further arms control steps. Meanwhile, domestic critics in the United States may not view a launch notification regime as a meaningful step toward the goal of limiting China’s nuclear forces. Ultimately, however, decisionmakers in each state should ask themselves whether the proposed regime would, in itself, enhance their state’s security; if it would, they should support it, even if they disagree about future steps. After all, U.S. participation would not reduce Washington’s leverage to push for further steps, and Chinese participation would not reduce Beijing’s ability to resist them.

Finally, there may be concern that notifications could cue additional espionage activities, such as pre-positioning intelligence assets, to monitor launches. This concern is likely to be most acute in Beijing because of its distrust of the United States and because it can currently avoid notification requirements, pursuant to its agreement with Russia, by launching away from Russia or by invoking an exemption permitted in “special cases.”7 However, even Moscow and Washington may have concerns about the increase in transparency compared to their existing bilateral regime.

That said, the warning afforded by launch notifications would not significantly enhance the effectiveness of intelligence-collection activities. China, Russia, and the United States already have, or are acquiring, early-warning satellites, which can continuously monitor launches. Similarly, visual reconnaissance satellites are likely to observe test preparations, even though the coverage they provide is episodic. Participants could try to take advantage of a launch notification by pre-positioning ships or aircraft. However, for safety reasons, tests over the ocean are generally preceded by safety warnings already. Meanwhile, aircraft are unlikely to be able to get close enough to monitor a test over land without violating other countries’ airspace.


1 Peter Vincent Pry, War Scare: Russia and America on the Nuclear Brink (Westport, CT: Praeger, 1999), 214–227.

2 Robert Jervis, Perception and Misperception in International Politics, new edition (Princeton, NJ: Princeton University Press, 2017), xxx-xxxviii, 143–172, 203–216.

3 “China, Russia Extend Notification Agreement for Ballistic Missile, Carrier Rocket Launches,” Xinhua, December 15, 2020,

4 In practice, safety warnings can act as de facto launch notifications. However, because China tests over land, it has no need to warn mariners and is inconsistent about warning aviators.

5 For a five-state proposal, see Frank O’Donnell, “Managing Nuclear Multipolarity: A Multilateral Missile Test Pre-Notification Agreement,” Washington Quarterly 43, no. 3 (2020): 177–196.

6 For example, “China Announces Missile Launch Training,” United Press International, July 18, 1995,

7 Agreement Between the Government of the Russian Federation and the Government of the People’s Republic of China on Notifications of Launches of Ballistic Missiles and Space Launch Vehicles, 2010, Section 2.5.