If nuclear power returns to space, it will find a lot of company. In the most recent tally provided by the Bulletin of Atomic Scientists, there are roughly 75 nuclear devices in space, 38 from the United States and 37 from Russia. Of these, 46 are in Earth orbit, 12 were left on the Moon or Mars, and 17 power deep-space probes.

And accidents have occurred.

In 1964, for example, an American satellite failed and re-entered Earth's atmosphere. As planned, it jettisoned its nuclear payload, releasing radiation over the Indian Ocean at an altitude of 75 miles, according to the Bulletin.

In 1973, the Apollo 13 spacecraft carried an RTG to be used to power a seismic station on the Moon. The mission was aborted and the spacecraft returned to Earth. The RTG was attached to the lunar module, which broke up on re-entry. NASA officials say the RTG re-entered intact, with no release of plutonium, and now sits on the floor of the Pacific Ocean.

In 1978, a Soviet radar reconnaissance satellite malfunctioned and crashed in Canada's Northwest Territory, releasing thousands of highly radioactive fragments into a lake and the surrounding area.

No evidence has tied these mishaps to any cancer cases or deaths.

Still, over the years, political and social pressure from these accidents, and others in terrestrial nuclear reactors, have combined to compel NASA to design Mars probes and rovers that rely on solar power.

But for robotic exploration, especially on the surface of a planet far from the Sun, with nighttime darkness and changing seasons thwarting solar collectors, nuclear power would be an indisputably more powerful exploration tool.

A stark example of solar power's shortcomings was provided by the successful Mars Pathfinder mission in 1997, which worked in tandem with the Sojourner rover to beam back pictures of the surface of Mars. While outlasting its 30-day life expectancy, the craft's batteries died just shy of three months after landing.

Researchers expected the batteries to die, because they required constant recharging from the solar panels. Solar energy cannot be used directly, because it fluctuates so much.

And solar panels are heavy, not to mention complicated to unfurl in space or on a planet.

Bob Anderson, a geologist and mission planner, said in a recent interview at JPL that the weight of solar panels and their poor performance compared to nuclear power severely constrain the amount of science that can be done for a given mission's price tag.

"Two things will kill a mission," Anderson says. "Power and mass."

And future Mars missions will require more of both. A pair of missions in 2003 will send the most advanced and capable rovers ever designed to study Martian geology and search for signs of water. If there, this water could provide the trail to any past life that might have existed on the Red Planet.

The craft may be sent inside giant craters, where orbiting spacecraft have spotted signs of water. But to ensure safety, the spacecraft will land in flat areas, likely near the crater center.

"But the best information is probably in the rim," Anderson says.

Anderson is helping engineers design rovers that will allow the geologist to remotely drill into rocks and figure out what they're made of. It is a critical science tool, but also a tremendously power-draining activity, he said.

Nuclear power could turn short, daytime-only missions into long, 24/7 operations, Anderson said. He notes, however, that rovers would have to be redesigned to make all their parts capable of sustaining such a long mission.

Naderi, the JPL manager, worries that Americans have been jaded into assuming that going to Mars is a relatively simple operation nowadays. But given that favorable planet alignments limit Mars missions to launching every 26 months, he laments solar-powered rovers die before the next one can be launched.

"People think [landing on Mars] is like driving to Grandma's on Sunday," Naderi said. "But it is expensive and it is horribly difficult to land on Mars. Once you do, you want to last more than 90 days."

While nuclear power can improve the efficiency of a rover, some say it is imperative for more ambitious missions.

An increasingly vocal group of space enthusiasts argues that the post-Apollo space program is stagnant due to the lack of a major goal. Many think that what's needed is a firm plan to set up permanent human colonies on the Moon or Mars.

Peter Eckart, of the Institute for Astronautics at the Munich University of Technology in Germany, says that if a lunar base is to be built anywhere except at the poles, where sunlight is constant, then "the only reasonable engineering solution is to go with nuclear power."

Likewise, others say, any future colonization of Mars will likely depend not just on nuclear electric propulsion, but nuclear power generation on the surface. Most engineers question whether even the most perfectly situated site can be sustained by solar power. And at best, these sites would not necessarily be located where researchers would want to explore.

Despite the benefits of nuclear power, Eckart is not one to discount the dangers.

"I'm personally not too much in favor of using nuclear power on Earth, if we can avoid it," he said after a recent conference on space colonization at Princeton University. "But in space, it's not a problem."

Eckart calls the fear of contaminating the lunar surface with radiation "total nonsense, because up in space there's so much radiation already -- all the galactic and cosmic radiation, all the stuff that's coming in from the Sun. A nuclear reactor does not make a difference at all. The only risk is launching it, and there you have to be careful from an engineering point of view."

Such a system would be launched in safer pieces, then assembled once at its destination, providing a further measure of safety, proponents say.