Ouch, yeah, I know ion drives are low thrust, but that takes quite a bit longer than I was thinking, and I didn't calculate exactly how long that would end up taking. (Although I get about half that time with those figures, 11 months for low Earth orbit->mars transfer. But slowing down at the far end even would be tricky and time consuming as well.)
Something that might be useful for moving cargo around, but yeah, not quite that suitable for manned missions unless you can bump up the thrust by a huge amount, which would require a lot of efficiency.
Well let's see, It's about 3.8 km/s from LEO to mars transfer orbit. and 6.1 km/s from low Earth orbit to low Mars orbit, if you're not areobreaking. For low earth -> low mars orbit, at 20000s (196 km/s exhaust velocity), that would take a mass ratio of around 1.0316, or around 2.37 tons of propellent for a 75 ton dry mass rocket. At 800s (7.84 km/s ), that would take a mass ratio of around 2.177, or around 88.3 tons of propellent for a 75 ton dry mass rocket. If you can areobrake at the mars end, and only need the low earth -> mars transfer costs, those numbers would drop to 1.47t and 46.8t.
So, you'd more than a 25t difference to account for propellent costs. If you're starting with a base 75t ship in both cases, If you can airbrake at the far end, it's at least 45t extra for NERVA, bringing it up to around 120t. If you have to slow down with rockets, it's more like an 86t difference, bringing it up to around 161t. And those are just for hoemann transfer orbits, minimum delta-v, maximum time.
So there's some potential there, if you can toss on an extra 30t of power production and engines (over and above the NERVA reactor mass, since I assumed a 75t dry mass for each so that would already be included for both), and get the thrust output up an extra order of magnitude. With something like http://en.wikipedia.org/wiki/Dual-Stage_4-Grid (only one I could find near 20k isp), 100N would take about 10MW (Ouch). According to http://en.wikipedia.org/wiki/Solar_panels_on_spacecraft You can get 300W/kg solar arrays, meaning you could get 10MW using about 33.3t (10MW at around Earth orbit side at least). That's almost practically viable (There would probably be extra mass in support structures and such, but in space, and under mm/s^2 scale acceleration, it wouldn't be under much stress) And you could save a bit of travel time in the middle part by adding extra propellent to do something a bit faster than a strict hoemann transfer orbit, with that much power, a bit of extra propellent can go a long way.