INADVERTENT BOMBING OF THE CHINESE EMBASSY
IN BELGRADE, YUGOSLAVIA, MAY 7, 1999
THURSDAY, JULY 22, 1999
Thank you for inviting me to participate in the hearing today on the
accidental bombing of the Chinese Embassy in Belgrade, Yugoslavia on
May 7, 1999 while the North Atlantic Treaty Organization (NATO) was
undertaking air operations against the Government of Yugoslavia in
response to their repression in Kosovo. I am also pleased to appear
with Director George Tenet, whose service to America I have admired
We are here today to discuss a painful subject -- the accidental
bombing of the Embassy of the People's Republic of China located in
Belgrade, Yugoslavia. War is a painful subject. War is about the
intentional destruction of things in order to impose a change of
policy of an enemy. We intended to bring destruction to the
infrastructure of violence and coercion inside Yugoslavia in order to
force the government of Yugoslavia to suspend its genocide against the
citizens of Kosovo. We intended to destroy those things -- and only
those things -- which the government of Yugoslavia used to conduct its
hostilities and repression. We never intended to destroy anything that
was not tied to Yugoslavia's war effort. Unfortunately, but not
uncommon in the conduct of hostilities, we made mistakes.
Director Tenet has testified before you earlier, and has restated
today, that the mistake that led to the unintended bombing of the
Chinese Embassy was an intelligence error. I don't see it that way. We
in the Defense Department have the responsibility for dropping bombs
and winning wars. Ultimately, the bombing of the Chinese Embassy was
our error. The intelligence community is indispensable to our work and
they did an absolutely splendid job helping the Department of Defense
during Operation Noble Anvil. I think our forces did a splendid job.
The results speak for themselves. We flew over 9,300 strike sorties
and attacked over 900 targets against a capable and sophisticated
opponent. We lost only two aircraft in these operations and
fortunately no NATO combatant was killed. But ultimately it was one of
our air crews executing one of our orders that led to this tragic
outcome. How did it happen? Why did it happen? Could it happen again?
How can we prevent accidents like this from happening in the future?
After Action Assessment
Secretary Cohen has directed that the Department undertake a
comprehensive assessment of our operations in Operation Noble Anvil
and he put General Joe Ralston and me in charge of that assessment. We
started this review only last month because the people who have to
undertake the review are also the same ones who fought the war and are
now enforcing the peace. Even though it is an extra burden just now,
we must undertake this review because the experience is fresh. The
data is still available.
The After Action Assessment will go into many more issues than the
Embassy bombing. We will be asking ourselves a wide range of hard
questions: Did fighting as a member of a coalition impose restrictions
that made it harder to achieve our military goals? How effective is
NATO, an organization designed for peacetime decision making, in the
fast-paced decision making of war operations? How effective were
Serbian deception procedures in blunting the effectiveness of our air
strikes? These are just examples of the hard questions we will ask of
ourselves in coming months.
The mistaken bombing of the Chinese embassy is only one of the
important questions in our review. We are looking at all of the
incidents of unintended damage during the air campaign. As I said, we
flew over 9,300 strike sorties and attacked over 900 targets, dropping
over 24,000 bombs or missiles. All together, we had 30 instances when
we caused damage we did not intend. As Director Tenet said, the degree
of accuracy is unprecedented in history. Never before has a military
force gone to war and exercised such care to minimize unintended
destruction and did so with such success. Where we expended enormous
effort to minimize unintended damage to innocent civilians, our
opponent fought a medieval war that attempted to maximize damage to
innocent civilians. That story is just now unfolding.
Of the 30 instances of unintended damage, one third were instances
where we damaged the target we wanted to destroy, but innocent
civilians were killed at the same time. You will recall the time one
of our electro-optically guided bombs homed in on a railroad bridge
just when a passenger train raced to the aim point. We never wanted to
destroy that train or kill its occupants. We did want to destroy the
bridge and we regret this accident. As I said, 10 of the 30 instances
of unintended damage fall in this category.
For the remaining 20 instances, 3 were caused by human error that
identified the wrong target, and two were caused by mechanical error
by our hardware. In 14 instances we have not yet determined whether
the unintended damage was caused by human error or mechanical failure.
We will determine that to the best of our ability during our after
The one remaining case of course is the most dramatic and it is the
subject of today's hearing. The bombing of the Chinese embassy was
unique in that we had a legitimate target that we wanted to hit; the
only problem is we had the target located in the wrong building. To my
knowledge, this is the only example of this failing in all of our
strike operations. Because it was such a tragedy, it merits special
review and attention.
Let me say that we didn't suspend critical judgment until the shooting
stopped. We knew we made a mistake in bombing the Chinese embassy
within hours. We started almost immediately to try to find out why it
happened and how to avoid similar mistakes in following air
operations. During the past two months we have refined our
understanding. Director Tenet outlined for you in his statement the
errors that led to mistaken bombing of the embassy. I don't intend to
repeat that in my testimony.
Three Key Questions
Instead I want to focus on the three key questions that are important
to Secretary Cohen and to any subsequent Secretary of Defense. Human
beings will always make mistakes. Where there are intentional acts, we
need to punish those responsible for such acts. We hold our men and
women in uniform to very high standards of performance. In this case,
however, I have reviewed the record carefully and I see no evidence of
intentional wrongdoing that led to this tragic accident. Where human
failings are unintentional and purely accidental, we face a different
challenge. Here we need to look at the systems and processes we impose
to guide and constrain individuals. Where did our systems and
procedures fail? Do our systems and procedures allow too many human
failings which can result in tragic outcomes?
You and I can't prevent human beings from making honest mistakes. I
and my colleagues at DoD are absolutely responsible for the systems
and procedures. If those systems and procedures permit failure or
exacerbate honest error, we are absolutely responsible for fixing
I believe there are three central questions that I must answer when
deliver our report back to Secretary Cohen.
First, is it too easy in our system to put things on a target list and
Second, can the Secretary of Defense trust the system used to
geolocate or "site" a target?
Third, is the current 'no strike" process adequate to prevent
unintentional damage in wartime?
In order to answer these three crucial questions, I must first begin
with a broad description of our process.
There are four aspects to the comprehensive targeting process. I will
use the following shorthand labels for these four sub-processes:
targeting guidance, target development, target approval, and
This is arguably the most important sub-process. These are the steps
we take when we receive from the President of the United States as the
Commander in Chief the specific national-level instructions on how to
fight the war, and what specifically he wants us to do and what he
wants us to avoid. This guidance is further developed into specific
instructions to battlefield commanders.
The President and his senior advisors decide, in broad strategic
terms, how to fight the war -- how long, with what intensity, and with
what limitations. It is in this sub-process that we decide what
actions a battlefield commander can undertake on his own authority and
what actions require specific approval from higher headquarters.
I have reviewed the record carefully and there is no evidence that
this part of the process contributed in any way to the accidental
bombing of the Chinese embassy. The guidance was clear and
unequivocal. No one had the authority to attack embassies and
everybody knew it.
This is the sub-process where various commands and offices review and
develop specific target "folders" for consideration by senior
commanders or civilian authorities. I must restrict what I say about
the specific details of this sub-process for security reasons. I can
say the following.
Targets are nominated and developed both at field headquarters and in
Washington in the national security establishment. The target
development process begins with a general proposition and results in a
comprehensive proposal -- known in our business as a target folder --
for approval in the approval process I will discuss shortly.
Targets developed in Washington are supposed to be evaluated by
multiple national intelligence agencies before they are submitted for
formal approval. As mentioned previously, targets may also be
developed in the field. Once approved by the CINC, any targets
requiring NCA approval for execution are submitted to the Joint Staff
for review and processing.
Targets that are developed in the field are submitted to the Joint
Staff for review before they are submitted for formal approval.
Targets developed in Washington are supposed to be evaluated by other
organizations in the national security establishment before they are
submitted for formal approval.
The crucial mistakes that were made -- as Director Tenet has testified
-- were made in this target development sub-process. Inappropriate
procedures were used to geo-locate a legitimate target. More
important, the subsequent review of the target folder in this target
development sub-process failed to catch the original mistake.
The third sub-process involves the review and approval of the target
folder. This is a crucial step because this is the phase in the
process when senior officers and individuals check to make sure the
target development process conforms to the target guidance. All the
important questions -- we thought -- were asked at this step: Is this
a legitimate target? How does it relate to our military goals? What
role does it play in our opponent's system of operations and how will
it affect him if it is destroyed? Can we constrain our intended damage
to this target only? What is the likelihood of unintended damage and
how can we minimize unintended damage by changing the time of day or
the physical direction of the attack?
My careful review of the process and the documentary evidence suggests
that the second most important mistake occurred in this phase. More
accurately, we didn't make a mistake by what we did in the approval
phase, but what we normally do not do. Normally we do not question the
mechanical aspects of target siting during the approval process. We
assume that the target has been properly located in geo-spatial terms.
And what we found was that in this case insufficient information was
provided in the target development sub-phase to enable senior
leadership to evaluate the physical siting decisions. I hasten to add,
however, that of the 900 targets struck, this is the only case of
misidentification during the target development process.
The fourth sub-phase is the execution phase, when our forces carry out
their assigned mission. In this case the bomber crew did exactly what
it was told to do. There is nothing in the execution phase that
contributed to the accidental bombing of the Chinese embassy.
Answering the Three Key Questions
With this background we can now answer the three key questions I posed
First, is it too easy in our system to put a target on the master
I think the evidence here is that our system is not lacking in this
regard. It is not too easy to get things targeted. We need to remember
that the intended target was an appropriate target. We wanted to
destroy the Headquarters of the Yugoslav Federal Directorate for
Supply and Procurement. This organization was directly involved in
Yugoslavia's war efforts and knocking it out would shorten the war and
limit our opponents ability to continue their violence. The evidence
suggests the target development and approval process appropriately
asked all the right questions about the FDSP. Unfortunately we thought
it was located in a building that turned out to be the Chinese
This leads me to the second question: Can the Secretary of Defense
trust the system and process we use to "site" targets?
Here it is clear that, in this particular case, the current system
failed us. It appears that the system worked well for 99.9 percent of
the targets, but when it fails spectacularly as it did here, we have
to say that the system must be improved.
There is always uncertainty in geo-locating targets. In some instances
the nature of the target simplifies the problem. Locating a railroad
switching yard or an oil refinery is relatively easy. They have a
unique character that is easily seen on a picture and their size makes
them relatively easy to locate on even the most general map. Geo-
locating a target when it is an office building in a major city is
dramatically harder and inherently less certain.
In this instance an analyst used an inappropriate procedure to
geolocate the target. The larger system failure is that the target
development process did not convey to other offices and individuals
the technique that was used in the siting process. Senior officers in
the approval process would have questioned the siting of the target
had the target folder indicated the method that was used by the
analyst. It did not. We have to fix that.
How we fix it is a matter of current review. Here we have to balance
the need for reasonable checks and balances with the imperative for
efficient operations in time of war. We will have specific
recommendations later this summer. At this stage I believe that one
possible remedy could be to include in the target folder some
description of the target siting methodology. The standards should be
higher for office buildings in urban settings than for large unique
structures such as oil refineries.
Third, is the current "no strike" system adequate to minimize
Here again I have to conclude that the answer is no. The system is
inadequate because it currently is dependent on data bases which are
not adequately updated. The concept of the no strike system is
fundamentally sound; the challenge lies in ensuring that the
underlying data bases on which the system depends contain, to the
maximum extent feasible, the most up-to date information available.
It is clear that no one wanted to attack the Chinese embassy. What is
embarrassing to me is that no one knew where the Chinese embassy
really was. There were a few individuals in our government Who knew
where it really was, but they were not consulted in the target
development process. But the bigger problem is that this episode has
revealed how insufficient our data base development and review process
is, given the extremely high standards that have evolved for modem
warfare American style.
Fifty years ago we knew we couldn't discriminate between embassies and
legitimate targets in a bombing campaign so we warned everyone
accordingly and pressed ahead. We can't do that today and don't need
to do that today. But that means we have to have data bases that are
sufficiently accurate to catch mistakes that will be made.
Let me use this as an opportunity to say something about the National
Imagery and Mapping Agency. There have been a lot of articles and
editorials in recent weeks that have blamed NIMA for this tragic
accident. NIMA is not at fault. The dedicated folks at NIMA did a
superb job supporting our war fighters. They are unsung heroes. The
fact is that an analyst inappropriately used a NIMA map developed for
a different purpose, and used an inappropriate technique to locate the
target on that map. The target development and approval process then
failed to catch this error through procedures that NIMA does not
control. I want to publicly thank LTG Jim King and the hard working
patriots at NIMA who worked so hard to support our war fighters in
this and every other operation.
Mr. Chairman, Congressman Dixon, let me conclude by saying that the
unintended bombing of the Chinese embassy was a tragic accident. This
review has led me to conclude that we could have avoided this accident
knowing what we now know. But when the accident was made, it is clear
that no one thought it was the Chinese embassy or wanted to hit the
Chinese embassy. This accident occurred because of a small but
important failure mode in our targeting system. It can be changed. It
needs to be changed. I pledge to you that it will be changed.