On the morning of September 11, 2001, after the second plane hit the World Trade Center and it was clear that the nation was under attack, US authorities issued an emergency alert, grounding air traffic and ordering nuclear power stations and other potential terrorist targets to go to their highest level of security. At the Three Mile Island nuclear plant in Pennsylvania, security guards sprang into action but soon ran into trouble: A gate designed to keep attackers out of the plant refused to close. Two hours later the guards were still struggling to shut it.

This was not the first time the Three Mile Island plant had given cause for alarm, of course. Twenty-five years ago, TMI suffered the most serious nuclear power accident in US history. In the pre-dawn darkness of Wednesday, March 28, 1979, the cooling system of TMI’s Unit Two reactor malfunctioned, sending temperatures inside soaring. For the next four days, the nation and the world watched with bated breath to see whether a full-scale core meltdown would follow.

That Friday morning, I was interviewing Leo Yochum, a senior executive vice president at the Westinghouse Electric Corporation, for a book I was writing about the nuclear industry. There had been no new reactor orders for five years, and now the accident at TMI was strengthening speculation that the US nuclear industry was doomed. But not at all, Yochum insisted. “I just don’t understand all this talk about nuclear being dead,” he complained. “There’s a nuclear imperative in this country. We know it, Wall Street knows it, and we’re prepared to meet it.”

Subsequent interviews with dozens of Yochum’s colleagues confirmed his point: The industry had no intention of giving up. Convinced of their technology’s historical inevitability, they would prepare for a comeback, redoubling safety efforts and launching a public relations campaign arguing that in the face of uncertain global energy supplies and the threat of climate change, nuclear power was actually the most secure and environmentally benign source of electricity available. If everything went as planned, the executives estimated, reactor orders should rebound by the 1990s.

That timetable has proven optimistic–there have been no new orders yet–but the industry is still here, still planning on a bright future and doing well enough financially in the meantime that it can afford to wait. Current revenues come mainly from servicing the existing 103 nuclear plants in the United States. Since those plants account for 20 percent of America’s electricity, there is no chance of shutting down the nuclear industry anytime soon. And over time, the lack of another major accident, along with the industry’s PR campaign, has increased public acceptance of new nukes to 50 percent of Americans.

The biggest obstacle today, as it was before the accident, is economics. The high capital costs of nuclear plants, which are caused mainly by their long and uncertain construction schedules (which in turn are driven by public concern over safety), leave them unable to compete for investment dollars. Coal, natural gas and improved energy efficiency are much less expensive.

But if Wall Street has turned its back on nuclear power, the industry’s longtime backers in Washington have not. The initial Bush-Cheney energy plan proposed building 1,000 new nuclear power plants by 2020–an average of one a week. Even the industry recognizes that goal as ludicrously unrealistic; it has called for approximately fifty new plants. And it has a strategy for making those plants cost-effective: slash construction schedules. Toward that end, the industry won approval from the Clinton Administration for new reactor designs that allow for quicker regulatory approval; now the Bush Administration is urging a “streamlined” regulatory process–i.e., one allowing the public less latitude to raise objections and cause delays.

But the legacy of September 11 may thwart these plans. The fourth plane hijacked that day crashed a mere 120 miles from Three Mile Island, highlighting the vulnerability of nuclear plants to terrorist attack. The industry rushed out a claim that government tests had shown nuclear plants withstanding a direct hit from a 757 jet. But the scientists at the Sandia National Laboratory who conducted those tests disavowed that conclusion. Meanwhile, other government tests had found that security forces at nuclear plants had failed to repel mock terrorist attacks more than 50 percent of the time–even though the forces knew well in advance exactly what day the “terrorists” were coming. Such shortcomings will surely raise concerns among any state or local public officials considering approval of new nuclear plants.

Nuclear waste, long the industry’s Achilles’ heel, poses similar difficulties, though the executives don’t understand why. “If you do a halfway decent job of disposing of it,” one top executive explained during my book interviews, “it’s at least a few hundred years before anything could go wrong, and [the citizens and legislators declaring their states off-limits to nuclear dumping] won’t even be here then.” At Bush’s urging, Congress in 2002 overwhelmingly approved Yucca Mountain, in Nevada, as a permanent disposal site for the nation’s nuclear waste. But the state of Nevada has promised an extended legal fight, and it will have plenty of evidence. A report by the nonpartisan General Accounting Office concluded that Yucca Mountain was chosen after “a failed scientific process.” And a member of the Nuclear Waste Technical Review Board, Paul Craig, recently resigned in order to release his warning that the current design of Yucca Mountain is certain to leak radioactive waste.

The industry and its government backers hope to overcome all these problems by tying nuclear power’s future to the new darling of alternative energy, hydrogen. Bush’s current energy bill proposes spending $1.1 billion to build a reactor in Idaho to try to produce hydrogen. Cindy Folkers of the Nuclear Information and Resource Service, an antinuclear advocacy group in Washington, criticizes this as both “unproven and absurd, since wind would be a much cheaper means of producing hydrogen.” Steven Kerekes of the Nuclear Energy Institute, the industry’s trade association, denies that the project is an industry subsidy, since the reactor will be owned by the Energy Department. Kerekes further argues that his industry is now simply making up for having received less federal support than renewable energy did under Clinton–a debatable claim that ignores how nuclear subsidies over the past fifty years have dwarfed those for renewable energy.

So, twenty-five years after Three Mile Island, nuclear power is still not dead in the United States. As the industry pushes for a revival, it has two advantages: US electricity demand is growing steadily as the Internet revolution advances; and, although most environmentalists won’t admit it, the industry has a point about the deadly effects of today’s dominant source of electricity production, coal. Coal smoke has killed millions of people over the past three centuries, including today, while its carbon content has pushed the climate toward a potentially catastrophic instability. Nuclear power may kill millions of people someday, but it hasn’t yet, and the longer the industry goes without another major accident, the more attractive nuclear will look–assuming, of course, that its costs are reduced. Better than either coal or nuclear would be a wholesale shift to a less centralized energy system based on solar and other renewables, using improved energy efficiency as the bridge to get there. But all these are political choices. The nuclear option will stay alive as long as the government keeps subsidizing it, and that seems likely as long as George W. Bush is President.