Now that I have your attention, I can tell you that it's pretty safe. But explosions like the picture on the left can happen. The photo is from the 2007 article "Blast Waves and Fireballs Generated by Hydrogen Fuel Tank Rupture During Fire Exposure" (Robert Zalosh, Firexplo). (Full Disclosure: Bob Zalosh has consulted for me on hydrogen safety issues at Giner Electrochemical Systems, LLC.)
The above black-and-white photo is an infrared image of the 7.7 m-diameter fireball that was taken 45 ms after rupture of a compressed hydrogen tank. This rupture was caused by putting a 265 kW propane burner directly under a hydrogen tank that had been installed in place of the fuel tank under an SUV. All protection normally provided by a pressure/thermal relief device had been disabled, in order to study the resultant explosion. The hydrogen tank failed catastrophically 12 min 18 seconds after the propane burner was ignited. This was obviously an aggressive test, and would likely only occur to a hydrogen-fueled vehicle if a gasoline leak ignited underneath. (I have not searched for documentation of the explosion resulting from a comparable experiment with a gasoline-fueled car, but I imagine the fireball could be bigger than the above hydrogen-fueled fireball!)
Results more favorable to the use of hydrogen as an automotive fuel have been provided by Michael Swain at the University of Miami, in the 2001 U.S Department of Energy-funded paper Fuel Leak Simulation.
Swain set fire to two cars, one fueled by gasoline and the other by hydrogen. The gasoline car had a 1/16" puncture in a fuel line. The hydrogen car had a leaking hydrogen connector.
The gasoline-filled car was completely engulfed in flames by 2 minutes 20 seconds, while the hydrogen car was essentially undamaged. Whereas the gasoline fire started as the result of a simple, small hole in the fuel line, for the hydrogen fire to occur, it would have taken the catastrophic failure of four separate safety systems, all at the same time, a highly unlikely occurrence.
What this blog post shows is that it is oftentimes difficult to compare the safety of alternative technologies. In some evaluations, sometimes "Technology A" is safer, while with other evaluations it turns out that "Technology B" is safer. Bottom line, hydrogen isn't so bad!