A methodology for supporting monetary risk assessment that combines probabilistic analysis with case-based reasoning is presented. Modern risk assessment requires the use of cost probabilities since for projects of reasonable complexity, deterministic methods for calculating costs are often misleading. Cost determination is complicated by the presence of dependent variables whose relationships with independent variables are project-specific. The probability distribution functions of independent variables and their relationships with dependent variables are constructed through an analysis of relevant past projects. These are used to compute the joint probability distribution functions using statistical techniques such as Monte-Carlo simulation and numerical integration. The methodology has been implemented in a software package called RACoPro and has been used in risk assessment of full-scale projects.