Atomic scale computer simulations are used to investigate the intracascade evolution of the defect populations produced in cascades in copper over macroscopic time scales. Starting with cascades generated using molecular dynamics, the diffusive transport and interactions of the defects are followed for hundreds of seconds in stochastic annealing simulations. The temperature dependencies of annihilation, clustering and free defect production are determined for individual cascades, especially including the effects of the subcascade structure of high energy cascades. The subcascade structure is simulated by closely spaced groups of lower energy MD cascades. The simulation results illustrate the strong influence of the defect configuration existing in the primary damage state on subsequent intracascade evolution. Other significant factors affecting the evolution of the defect distribution are the large differences in mobility and stability of vacancy and interstitial defects and the rapid one-dimensional diffusion of small, glissile interstitial clusters produced directly in cascades. Annealing simulations are also performed on high-energy, subcascade-producing cascades generated with the binary collision approximation and calibrated to MD results.