Its a rigid foam that must be cut/sanded. It does not hot-wire and attempts to cut it that way generate toxic fumes. It is sold in sheets of various thicknesses from 1mm up to (at least) 6mm. The number after the name is its density in Kg / cubic metre. The stuff is rigid to handle and very good in compression. I have been told that it can absorb moisture and warp in some circumstances. Tom Koster had a ribless F1C back in '85 with all surfaces formed from moulded glass/Rohacell/glass sandwiches and he said he'd had warping problems without the inner glass skin. However Tom is a lawyer by training and has built structures in the past that I would not have bothered with and Jim Bradley used naked Rohacell ribs with carbon caps for a long time without problems.
By comparison with Balsa its very light. 1 Kg.m^3 is 0.062427969 lb/cu.ft., so Rohacell 31, 51, 71 are respectively 1.935267035, 3.183826412, 4.43238579 lb cu/ft. Martyn Cowley once told me that Rohacell 71 is good in F1A inner spars, so it must be at least as good in compression as end grain medium balsa, and that Rohacell 51 is good in all other spars.
You can work it with modelling tools and adhere it with epoxy and (I think) cyano. It seems that a good way to make D-boxes for Coupe and F1J tail surfaces is to vacuum-bag a layer of 83 gsm carbon onto 1.5 or 2 mm Rohacell 31 sheet and use this for a full depth spar and closure for a shell made from a single layer of 83 gsm carbon cloth. This assembly forms a large, hollow LE that replaces both the conventional LE and the main spar.