Probabilistic analysis is increasingly recognized as the standard for ecological risk assessment. Unlike deterministic approaches where a model outcome is always the same, probabilistic analysis incorporates real-world uncertainty into a model, due to estimated variability in model inputs, as well as random variability (stochasticity). We use techniques such as distribution-fitting and Monte Carlo analysis to evaluate uncertainties of numerical models brought about by uncertainty in input data, model parameters, spatio-temporal scale, model assumptions and model structure. Samples of model inputs are drawn from probability distributions either locally (one-at-at-time sampling), or globally, to create a matrix of scenarios or possibilities that model outputs can take. The full spectrum of model realization becomes the basis for risk assessment and managerial options.