Indeed - a catastrophic launch failure would produce pretty widespread contamination. Perhaps you could mitigate this by launching from somewhere in the middle of the ocean, but in all honesty I don't think even that would get past the huge political hurdles.<br /><br />NTR engines can still be extremely useful even if not used for launch. The amount of dV required to go to the Moon or Mars is about 40% of that used to get to LEO, so using a nuclear rocket for translunar or trans-Mars injection will easily let you get nearly twice the payload (assuming dV=5000 for TLI, braking into lunar orbit and TEI, this is about what Apollo used) chem Isp = 421 (S-IVB), nuke Isp=810 (NERVA), rocket fueled mass = 150t, chem payload = 45t vs. nuke payload = 81t ) to your destination compared to using chemicals all the way.<br />The ratio of payload for nuclear vs. chemical is not affected by increasing the fuelled mass of the rocket, but it IS affected by the delta-V - the more delta-V, the better the nuclear performs). <br /><br />You get even bigger benefits by using the nuclear rocket to brake into lunar or martian orbit, and some of the designs also provide for the use of the engine as an electrical generator when not in use for propulsion.<br /><br />With a nuclear rocket, it MIGHT even make sense to brake back into LEO - particularly since the treehuggers wouldn't like the hot core to re-enter. Perhaps for safety and mass-efficiency the capsule could detach prior to perigee and use a small thruster to put itself on a re-entry trajectory, while the nuke propulsion stage could stay on an orbit with a perigee out of the atmosphere, then burn to brake into orbit ready for the next trip.<br /><br />As an aside, the rocket needed for TLI etc would be much smaller than one used as the first or second stage of a launch vehicle. In fact the NERVA program already tested engines of comparable size, so they are definitely feasible, and much cheaper to develop than a monster first stage. <b></b>