The latest generation of parallel supercomputers promises to deliver computational performance that is orders of magnitude higher than that of the sequential supercomputers. The major obstacle to usingthese supercomputers for reservoir simulations is the lack of software. Most existing reservoir simulators were developed to run on sequential computers and must be restructured to take advantage of these parallel systems.The major bottleneck in the parallelization of a reservoir simulator is the solution of the linearized finite difference equation. Currently, the most robust techniques for solving large scale reservoir simulation problems are preconditioned conjugate gradient like methods. However, the more robust preconditioners such as nested factorization and ILU factorization are not suitable for parallel computers. This paper describes a parallel linear equation solver that is suitable for large and small scale parallel computers. It is based on the preconditioned generalized conjugate residual method. The new preconditioner is a parallel variant of nested factorization. For the problems tested, the new solver is as robust as the more traditional solvers for sequential computers. Results are presented for large scale reservoir models running on the Cray C-90 and a massively parallel CM-2.