The Pentagon's Missile Defense Agency (MDA) recently admitted that it was pushing back plans to put up a space-based missile defense test bed to at least 2008. But that does not mean the agency has given up on developing orbiting interceptors for shooting down enemy missiles in their boost-phase, shortly after their launch.

MDA officials, and hawkish proponents of using space for missile defense, continue to assert that it is technically feasible to design such a system using only 300 to 600 interceptors and costing $50 billion.

A recent study by an illustrious panel of physicists begs to differ. Even though they themselves admit to using "extremely optimistic" technical parameters, the American Physical Society (APS) in a July 16 study found that a bare-minimum system would require at least 1,600 missiles. Such a limited system would be able to defend only the continental United States (not including Alaska) and be able to shoot down only one solid-fuel ICBM coming in from North Korea (the sort the Pentagon predicts Pyongyang and other countries are likely to have within 10 to 15 years).

And the U.S. interceptors would have to be substantially larger and faster than ever built before, not to mention larger and faster than currently estimated by MDA. All totaled, the interceptors would weigh 2,000 metric tons.

While the study, "Report of the American Physical Society Study Group on Boost-Phase Intercept Systems for National Missile Defense: Scientific and Technical Issues," did not provide any cost analysis, doing the math is fairly simple. Average launch costs have hovered for decades at about $22,000 per kilogram. A metric ton equals 1,000 kilograms. So, this best-case scenario for space-based missile defenses would cost $44 billion just to get the interceptors into orbit.

Some experts argue that, given the volume of space launches that would be required to boost the system, launch costs could conceivably over time come down to half that per kilogram sum: $11,000. If this is true, then such a system could be put into orbit for only $22 billion.

But here's the rub: The physicists themselves admit that the system described above is based on assumptions that are optimistic enough to border on unrealistic. Under more realistic technical parameters, a system to defend the continental United States against a North Korean launch would involve 3,600 orbiting interceptors, at a cost of either $99 billion, or using the lower launch cost figure, $49.5 billion. However, the study itself notes that even these "more realistic" assumptions are quite optimistic, not only in pushing the edge of what is technically feasible but also in that the space-based system described is one in which every element works perfectly 100 percent of the time --something unheard of in the annuals of U.S. weapons development.

There is more bad news. To cover Alaska, more than double the number of interceptors would be required to defend against a North Korean ICBM, thus more than doubling the cost (more than $198 billion or more than $99 billion).

To defend against a single shot from Iran (another of the countries labeled by U.S. President George W. Bush as part of the axis of evil, and a country with a ballistic missile program), the study found, is more difficult and would require more interceptors. The study found under its more realistic scenario, that 5,700 interceptors would be required, weighing 7,000 metric tons, equaling a launch cost of $154 billion (or $77 billion).

Some might say that such price-tags are not out of line for a future strategic system, given what the United States has spent on its nuclear arsenal. That may be so. But remember, these figures involve only the direct cost of launching the space-based interceptors. Such interceptors, which according to the study must be much faster and much larger than any to date, would have to be developed and built. More cost. In addition, a complex computerized system to control the interceptors would have to be developed. Yet more cost. Finally, a sophisticated new system of detecting, tracking and targeting ICBM launches and nearly instantaneously providing that data to the orbiting interceptors, would be required. Substantially more cost.

Even more troubling is the fact that the study's more realistic scenarios include assumptions that are forgiving in the extreme. For example, these scenarios include only 30 seconds of time for a decision to fire - the best-case analysis assumed an automatic shot once a potential target was detected. This is highly problematic, in that it is impossible to tell during the early boost-phase whether what just went up was an ICBM or a space-launch vehicle carrying a satellite (or, in the case of China, possibly astronauts). To put it mildly, it seems unlikely that any U.S. commander in chief would be comfortable with automating such a momentous decision.

Furthermore, as noted above, these scenarios all are based on essentially a one-shot (in some cases, two-shots), one-kill architecture. This means there is no margin for error; no redundancy in the system. If North Korea decided to launch two ICBMs (once they get them) at Alaska from nearby launch sites, the U.S. networks postulated by the study would most likely be useless. To be able to target multiple interceptors at each incoming ICBM, however, not only involves even more astronomical costs, but also raises the technical problem of ensuring that the interceptors don't become confused and mistake another of their fellow interceptors for the target.

The APS study, in its generosity, called space-based missile defense "impractical." A more realistic look at the data shows that it is wildly so.

Theresa Hitchens is vice president and director of the Space Security Project at the Center for Defense Information. This article originally appeared as an opinion piece in Space News on July 21.