Submarine accretionary prisms are composed largely of poorly consolidated sediments, particularly near the toe. Field and laboratory observations, by Dan Karig in particular, have shown that these sediments can display both brittle and ductile behavior, depending on local stress conditions, stress history, and physical properties of the sediment. This complexity in deformation behavior of accreted sediments has stymied many attempts to develop numerical models for the study of accretionary processes; continuum models may capture the overall geometry of the prism, but cannot reproduce the complex structure and evolution of natural prisms. A numerical technique known as the distinct element method (DEM) provides a way to simulate accretionary prisms as discontinuous systems, e.g., assemblages of particles that interact individually with eachother to generate the behavior of the whole. Because particles obey simple physical laws of interaction, the technique defines rather than relies on the constitutive behavior of the assemblage. In this "numerical sandbox", it becomes possible to explore relationships among sediment properties, local stress conditions, deformation mode, and prism structure. As a test of the feasibility of this technique, a series of DEM simulations were conducted using several thousand particles within compressional boundaries. These simulations qualitatively reproduce the behavior and geometries of natural prisms. Depending on material strength, the prism grows by smooth advance of the deformation front (ductile deformation) or by formation and propagation of discrete frontal thrust faults (brittle deformation); broad folds form above thrust faults; out-of-sequence thrusts displace faults and strata within the prism. Estimates of prism taper show variations with internal and basal friction in approximate agreement with critical Coulomb wedge theory.