The multigroup diffusion equations are solved by treating them as an initial value problem. The inherent error growth is controlled by repeated conditioning transformations; the error bounds on the final solution are set by the frequency of conditioning. The stabilized march technique (SMT) is comparable in speed to AIM-5 for problems involving downscatter only. The SMT is shown to be relatively insensitive to the type of scatter matrix involved and, hence, presents an advantage for problems with full scatter matrices. The technique is readily adaptable to flux synthesis, and an example is given for expanding the thermal flux in Laguerre polynomials. The SMT performs equally well in calculating higher order eigenvalues and eigenfunctions.