A multiscale computational framework is presented for modeling and simulation of point-defect aggregation in crystalline silicon. Large-scale molecular dynamics simulations based on empirical potentials are employed to calculate both parametric and mechanistic data, which are passed onto lattice kinetic Monte Carlo and continuum rate equation models. Multiple model predictions are compared to experimental data and are shown to provide an accurate, comprehensive picture of vacancy aggregation. (C) 2007 Elsevier B.V. All rights reserved.