The decision came with less than 2 1/2 hours to go before launch, as the seven astronauts were almost done boarding the spacecraft. Up until then, rain and thunder over the launch site appeared to be the only obstacle to an on-time liftoff.
The same baffling problem cropped up during a launch pad test back in April, and NASA has been struggling ever since to figure out the source of the trouble. But the topic came up repeatedly at meetings of top-level NASA managers this week, and the space agency said that it believed it had worked around the problem by replacing cables and other electronics aboard the shuttle.
NASA officials said it's not clear whether the problem is with the sensor itself or with the cables or the electronics boxes aboard Discovery.
It was also unclear whether the shuttle could be fixed at the launch pad or would have to be rolled back to the hangar, which would mean a much longer delay.
NASA deputy shuttle program manager Wayne Hale said that even in a best-case scenario, the shuttle will not be ready for launch before Saturday.
As recently as Monday, Hale described the sensor problem seen in April as simply an "unexplained anomaly."
The back-to-back failures suggest the possibility of a wider problem than one or two bad pieces of equipment.
NASA has until the end of July to launch Discovery, after which it will have to wait until September — a schedule dictated by both the position of the international space station and NASA's desire to hold a daylight liftoff in order to photograph the shuttle during its climb to orbit.
"This is a critical system. They have four of these sensors, and they require all four to be operating at launch. Right now, it looks like they've got problems with two of those," said CBS News Space Consultant Bill Harwood. "It's a big disappointment."
The sensors let the computers know when the tank is dry and they can shut the engine down. If the engines were to continue running at high speed without fuel, it could lead to a catastrophe.
NASA said it appeared that the sensor was showing a low fuel level, even though the tank was full with 535,000 gallons of super-cold liquid hydrogen and oxygen.
The sensors "for some reason did not behave today and so we're going to have to scrub this launch attempt," launch director Mike Leinbach told his team. "So appreciate all we've been through together, but this one is not going to result in a launch attempt today."
During a fueling test of Discovery's original tank in April, one of its sensors gave intermittent readings. NASA could not figure out the exact reason for the failure but replaced the entire tank anyway to install a heater to prevent a dangerous ice buildup.
Shuttle managers considered conducting a fueling test at the launch pad on the replacement tank, but ruled it out to save time, saying that the actual fueling on launch day would be the ultimate test.