Models are essential to synthesize knowledge on weeds and to design integrated weed-management strategies. These models must rank cropping systems as a function of weed infestation, and account for variability in effects to estimate probabilities of success or failure. Three case studies are presented: (1) an empirical static single-equation model that directly relates weed biomass to crop management, with few inputs and parameters, (2) a matrix-based multiannual model predicting a few key weed stages annually, from weed control options and a few parameters, (3) a mechanistic process-based multiannual model predicting detailed soil, crop and weed state variables daily, with an individual-based 3D canopy representation, requiring hundreds of inputs and parameters. The chapter concludes that models using a mechanistic representation of the cropping-system ´ environment interactions are best for quantifying effects and their variability, combined with a subsequent transformation with in silico experiments into empirical models of key cropping-system components.