DCI Statement on the Belgrade Chinese Embassy Bombing
House Permanent Select Committee on Intelligence
22 July 1999
Mr. Chairman, Dr. Hamre and I are here today to explain how a series
of errors led to the unintended bombing of the Chinese Embassy in
Belgrade on May 7th.
We will try to describe to the best of our ability -- in this open,
public session -- the causes of what can only be described as a tragic
mistake. It was a major error. I cannot minimize the significance of
this. The ultimate responsibility for the role of intelligence in this
tragedy is mine. I've told my own people that we will not hide behind
excuses such as stretched resources or time pressures. It is precisely
when the pressure is intense -- life or death decisions are being made
-- that the President and the American public expect us to provide the
best intelligence in the world. Clearly, in this case we failed to do
But before we tell how this happened, I think it is important to
provide some perspective. Dr. Hamre will tell you that the United
States and our allies flew thousands of sorties and struck many
hundreds of targets over 78 days with very few errors. America's
success, in this as in previous conflicts, owes much to the
extraordinary work of our intelligence services. The specifics of our
contributions cannot be made public Mr. Chairman, but as this
committee knows, we provided our forces detailed knowledge of the
enemy, his intentions, his dispositions, and his weapons.
Mr. Chairman, the nature of warfare has changed. When cities were
struck in past wars, none doubted that civilians, embassies,
hospitals, and schools would be in harm's way. Today, our ability to
strike precisely has created the impression that sensitive sites can
be safe in the middle of a war zone. Our desire to protect innocents
in the line of fire has added an enormous burden on all of us that we
accept. It is our job to do our best to ensure that only appropriate
targets be struck.
I think it is useful to note that this episode is unusual because the
CIA does not normally assemble, on its own, target nomination packages
containing the coordinates of specific installations or buildings. The
targeting support typically provided by CIA is usually at the
strategic and planning level, such as analytical judgments on the
kinds of targets that are the most important, commentary or specific
information concerning targets selected by the military or others, and
information that assists the military in identifying future targets.
In addition to describing how this mistake happened, I will also
outline the corrective actions that we are taking within the
government to ensure that -- as far as humanly possible -- that there
is no repeat of this type of incident.
The attack was a mistake. Let me emphasize, our investigation has
determined that no one -- I repeat no one -- knowingly targeted the
Chinese Embassy. Speculation to the contrary is simply unfounded. No
one, at any stage in the process, realized that our bombs were aimed
at the Chinese Embassy.
There were three basic failures. First, the technique used to locate
the intended target -- the headquarters of the Yugoslav Federal
Directorate for Supply and Procurement (FDSP) -- was severely flawed.
Second, none of the military or intelligence databases used to
validate targets contained the correct location of the Chinese
Embassy. Third, nowhere in the target review process was either of the
first two mistakes detected.
The unintended attack happened because a number of systems and
procedures that are used to identify and verify potential targets did
Chronology of Events
To help understand the circumstances which led to the mistaken
bombing, let me offer a brief chronology of events.
In March of this year, U.S. intelligence officers began considering
the FDSP headquarters as a potential target for NATO ALLIED FORCE
strike operations. The FDSP was a legitimate target given its role in
support of the Yugoslav military effort.
We had the street address of the FDSP headquarters as "Bulevar
Umetnosti 2" in New Belgrade. But military forces require precise
geographic coordinates to conduct an attack. During a mid-April
work-up of the target, three maps were used in an attempt to
physically locate the address of the FDSP headquarters: two local
commercial maps from 1989 and 1996, and one U.S. government map
produced in 1997. None of these maps used had any reference to the
FDSP building. None accurately identified the current location of the
Please keep in mind that the location of the Chinese Embassy was not a
question that anyone reasonably would have asked when assembling this
particular target package. This package was intended to strike the
FDSP headquarters and nowhere else.
In an effort to pinpoint the location of the FDSP building at Bulevar
Umetnosti 2, an intelligence officer used land navigation techniques
taught by the U.S. military to locate distant or inaccessible points
or objects. These techniques are known as "intersection" and
"resection." They can be used for general geolocation, but should not
be used for aerial targeting because they provide only an approximate
location. Using this process, the individual mistakenly determined
that the building which we now know to be the Chinese Embassy was the
FDSP headquarters. The true location of the FDSP headquarters was some
300 meters away from the Chinese Embassy. This flaw in the address
location process went undetected by all the others who evaluated the
FDSP headquarters as a military target.
A critical lesson that emerges from this event is that particularly
when providing targeting nominations in urban areas, it is important
to provide an accurate appreciation of our confidence in the location
of a target, and the evidentiary basis for how that location was
The incorrect location of the FDSP building was then fed into several
U.S. databases to determine whether any diplomatic or other facilities
off-limits to targeting were nearby. We try to avoid damage to
sensitive facilities like embassies, hospitals, schools and places of
worship and look to see what risk to them a nearby strike might pose.
Moreover, satellite imagery of the target provided no indication that
the building was an embassy-no flags, no seals, no clear markings.
Location of Chinese Embassy
Multiple databases within the Intelligence Community and the
Department of Defense all reflected the Embassy in its pre-1996
location in Belgrade. Despite the fact that U.S. officials had visited
the Embassy on a number of occasions in recent years, the new location
was never entered into intelligence or military targeting databases.
If the databases had accurately located the Chinese Embassy, the
misidentification of the FDSP building would have been recognized and
Why wasn't the Chinese embassy correctly located? It is important to
understand that our ability to locate fixed-targets is no better than
the data bases, and the data bases in this case were wrong. Further,
it is difficult-actually it is impossible--to keep current databases
for cities around the globe. The data bases are constructed to catalog
targets not non-targets. In general, diplomatic facilities--our own
being an exception because of the need to plan for evacuation--are
given relatively little attention in our data bases because such
facilities are not targets. Military targets are the top priority
because of the danger they pose to our own forces.
In this context I would add my belief that too much public emphasis
has been given to the fact that the 1997 U.S. Government map did not
reflect that the Chinese Embassy had moved. This criticism overstates
the importance of the map itself in the analytic process. Maps of
urban areas will be out of date the day after they are published. What
is critical is having accurate data bases.
We have subsequently found maps which show the correct current
location of the Chinese Embassy although there are others, including
some produced after 1996 by the Yugoslav government, which do not.
Some of our employees knew the location of the Chinese embassy. But
keep in mind that we were not looking for it. None of these
individuals was consulted as the target was selected and reviewed and,
as a result, we lost the opportunity to learn that the building
targeted was not the FDSP headquarters. We have also found one report
from 1997 that gave the correct address of the Chinese Embassy but
that information was ancillary to the focus of the report and
unfortunately the address was not entered into the data base.
Late Concerns About the Target
Very late in the process, questions were raised by an intelligence
officer as to whether the building targeted was in fact the FDSP
headquarters or might be some other unidentified building. At no time
was there any suspicion that the building might be an Embassy. This
officer had become aware of the nomination by chance, and remembered
having seen information a few years earlier that the FDSP building was
located a block away from the location identified. Although the matter
had nothing to do with his usual responsibilities, this officer
registered his concern and sought to clarify the facts by contacting,
or attempting to contact, other working level officials who were
involved in the preparation of the nomination package.
On Tuesday, 4 May--three days before the bombing occurred--this
officer telephoned an officer responsible for the target at the Joint
Task Force NOBLE ANVIL in Naples. He told him that he believed the
FDSP headquarters building was a block away from the identified
location and said that he was trying to resolve this discrepancy. That
evening, he obtained information that tended to confirm his belief
that the building had been mislocated. Due to a variety of
circumstances, this officer was unable to relay this information
before departing for training 6-7 May. At that time this officer was
unaware that the FDSP headquarters was on the target list.
On his return to his office on Friday afternoon, 7 May, the officer
learned to his surprise that the FDSP building was on the target list
for bombing that night. He attempted without success to re-contact the
officer in Naples with whom he had spoken earlier in the week
concerning the "discrepancy." He raised his concerns with another
officer at Naples and learned that the aircraft was already en route
to the target. He tried to convey his concern that the building
targeted may not have been the FDSP headquarters. Those in Europe
state that they believed that he was trying to convey that while it
might not be the FDSP headquarters, it was still a legitimate FDSP
target. While recollections differ of exactly what was said and what
was heard, there is no doubt that no one knew that the facility in
question was an Embassy. The strike took place shortly thereafter.
Throughout this series of missed opportunities, the problem of
identification was not brought to the attention of the senior managers
who may have been able to intervene in time to prevent the strike.
What Went Wrong?
At this point, I would like to identify the principal shortcomings
that caused this accident to take place.
First, the approach used to determine the location of the FDSP
headquarters was inappropriate for targeting. There were three
meetings at CIA that reviewed the target nomination. The method of
identification was not briefed, questioned, or reviewed. Therefore,
the initial misidentification took on the mantle of fact. The absence
of discussions on this matter resulted in a target package that
contained no cautionary language on the location of the FDSP
headquarters. Absent cautionary language, reviewers at EUCOM and the
Joint Staff mistakenly assumed the location was accurate. This made it
unlikely that they would focus on the need to re-validate the target's
Second, within CIA there were no procedural guidelines for the
officers involved in targeting to follow, and there was little senior
management involvement in guiding the targeting process. Although our
military support organization had been involved in targeting matters,
they had not previously been involved in the approval of target
nomination packages unilaterally proposed and wholly assembled at CIA.
This occasion was precedent-setting.
No institutional process existed within CIA for ensuring that all
resources were brought to bear on the FDSP nomination.
Third, reviewing elements at EUCOM and in the Joint Staff did not
uncover either the inaccurate location of the FDSP headquarters or the
correct location of the Chinese Embassy was the result of both data
base shortcomings and procedural errors. The data base reviews were
limited to validating the target data sheet geographic coordinates
with the information put into the data base by the NIMA analyst. Such
a circular process did not uncover the original error and made us
susceptible to a single point of data base failure. While collateral
damage assessments were performed and indicated there were no
sensitive facilities in the area, these assessments were based on
incomplete data on the location of those sensitive facilities.
Individuals in both CIA and the DoD who knew the correct location of
the Chinese Embassy should have been consulted.
Fourth, the critical linchpin for both the error in identification of
the building and the failure of the review mechanisms is the
inadequacy of the supporting data bases and the mistaken assumption
the information they contained would be necessarily accurate. The
misidentification of the targeted building as the FDSP headquarters
would not have occurred had the data bases had the correct location of
the Chinese Embassy. All the data bases that contained information on
the Chinese Embassy placed it at its original, pre-1996 location some
four miles away. Thus, the question of possible damage to the Embassy
was never a consideration.
U.S. officials who had served in Belgrade were aware that the Chinese
Embassy had moved sometime in 1996. The information, however, was not
entered into the data bases we rely on for our targeting and mapping.
In this context I would add my belief that too much public emphasis
has been given to the fact that the 1997 NIMA map did not reflect that
the Chinese Embassy had moved. This criticism overstates the
importance of the map itself in our analytic process. Maps of urban
areas will be out of date the day after they are published. What is
critical is having accurate data bases.
Data base maintenance is one of the basic elements of our intelligence
effort, but it is also one that has suffered in recent years as our
workforce has been spread thin. Some have suggested that this failure
is the consequence of resource shortfalls. A more fundamental problem
is not the absolute level of resources, but the application of
resources at our disposal. We have diverted resources and attention
away from basic intelligence and data base maintenance to support
current operations for too long.
Data base production and maintenance has been routinely accorded a low
priority and often overlooked in production planning and scheduling.
Data base production is often the first activity curtailed when
resources are tight. Data base production is widely viewed as low
visibility, unrewarding, and unappreciated. Leadership attention and
emphasis on data base production is infrequent, episodic, and
Corrective Actions Taken
We are continuing our in-depth review of this tragic incident. Based
on our initial findings, it is clear that this mistake occurred not
because of just one organization, or because of one individual.
Nevertheless, I am evaluating our performance in this instance to
assign individual responsibility and identify procedural reforms.
Our goal is to ensure that such a mistake does not happen again. To
this end, we are implementing corrections to prevent such mistakes in
In addition, the following near-term corrective actions are already
DIA and NIMA have established rapid response procedures for critical
database updates. We are strengthening our internal mechanisms and
procedures for selecting and validating targets and we are increasing
the priority placed on keeping databases current. The Community and
other government agencies will explicitly report whenever foreign
embassies move or are built. This information will then be forwarded
and incorporated into our intelligence and military databases. In
future conflicts, we will contact other governments to help identify
and locate their facilities.
Experience tells us that humans err. Knowing that, we constructed
elaborate procedures to check and double-check our work. In this
specific case, the checks and balances failed. The President of the
United States has expressed our sincere regret at the loss of life in
this tragic incident and has offered our condolences to the Chinese
people and especially to the families of those who lost their lives in
this mistaken attack.