To understand why the seemingly benign proposal could lead to such dire unintended consequences, it helps to appreciate how India's nuclear efforts came to be in such a troubled state, starting with the unique history of the program. In 1954, seven years after the country became independent from Britain, the newly founded Department of Atomic Energy laid out a three-stage plan for nuclear power. The first phase was to construct and operate heavy-water reactors fueled by natural uranium and then reprocess spent fuel to separate plutonium, which could be recycled as new reactor fuel. But plutonium can also be used in weapons—a fact the DAE remained largely silent about. The second and third phases involved developing breeder reactors that could use thorium. Unlike the uranium-235 contained in natural uranium, thorium can't undergo fission at low energies. However, thorium is fertile; when bombarded with neutrons, it can be transformed into uranium-233, which can be used to fuel a reactor.

All of India's 17 existing reactors are part of the first phase, as are five of the six facilities under construction. Efforts to use thorium as a fuel have not gone far. An experimental breeder, the Fast Breeder Test Reactor at Kalpakkam, in southern India, constitutes the only functional part of the second phase. The Fast Breeder Test Reactor has been continually plagued with problems since its start-up in 1985. One glitch within the reactor vessel during a fuel-transfer operation in May 1987 forced the reactor to shut down for two years to undergo repairs. It wasn't until 2000 that the reactor managed to run continuously for more than 50 days.

In October 2004, after 20 years of planning and repeated delays, the DAE finally began building India's first industrial-scale fast breeder reactor. But it isn't expected to be completed until 2010—and only if all goes according to plan. Exorbitant costs, safety concerns, and engineering problems have effectively killed similar breeder programs in the United States, France, and Germany and set them back severely in Japan and Russia.

Even by the Indian government's optimistic timetable, breeders using uranium-233 as fuel won't start operating until the middle of this century, decades behind schedule. And nuclear power is unlikely to contribute even 10 percent of India's electricity generation in the next few decades.

Despite such shortfalls, the DAE continues to pursue its three-phase plan, and in particular the long-standing goal of developing reactors that use thorium and uranium-233. There is some logic to preferring thorium: by the International Atomic Energy Agency's estimates, India has about 225 000 metric tons of thorium, or almost a third of the world's reserves. Uranium reserves, by contrast, are more limited and of poor quality in India.

Under the circumstances, India will be compelled to rely primarily on uranium for some decades. Assuming that its reactors run at 75 percent capacity on average, India needs about 510 metric tons of uranium a year; recent figures suggest its annual production is about 200 metric tons. The DAE has been making up for some of the shortfall by using uranium stockpiled when India's nuclear generating capacity was much smaller.

Even so, reactor electricity outputs, as reported on the Nuclear Power Corp. of India Web site, fell to 56 percent in 2006–2007, down from 74 percent just three years earlier. With new reactors coming online and without uranium imports, this decline will continue. The department's best efforts to open new uranium mines and a new uranium ore concentrating plant have meanwhile met with stiff resistance from local communities, primarily because of the many health problems that existing mines have created, such as well-documented increases in birth-defect rates in nearby villages.

The DAE also maintains that nuclear power is much cheaper than other sources of energy, but the agency's budgets paint a different picture. When the DAE has compared the costs of generating electricity from nuclear reactors with those from coal-fired plants, India's staple source of electricity, it has assumed that the coal plants are located far from the source of their coal. Obviously, that assumption inflates coal transportation costs and biases the comparison in favor of nuclear power.

In the early years, the DAE claimed that nuclear power was cheaper than thermal plants that were located more than 600 kilometers away from the coal mines. By the 1980s the distance had crept up to 800 km. A 1999 study increased the distance to 1200 km. In fact, though, one-third of India's coal plants are located right next to a mine pithead, and another quarter or more are within 500 km of one.

Even the 1200-km claim does not hold up to scrutiny. Two researchers from the International Energy Initiative and I compared the cost of producing electricity at India's most recently commissioned nuclear reactor, a 220-megawatt heavy-water reactor at the Kaiga atomic power station in the southern state of Karnataka, with electricity from a nearby coal plant that is 1400 km away from a mine. We found that the nuclear plant was about 8 percent more expensive at the government-determined rate of return on investment, which reflects the present value of future benefits and costs. At market rates of ROI, however, it could be 50 percent more expensive.