The objective of this paper is to establish a method for the early forecasting of product realisation costs. A detailed explanation of the key terms and of the general topic is followed by a definition of the decisive requirements of a method for cost forecasting. The status that is currently published in the relevant specialist literature is examined and assessed. This study shows that the group of the multivariate parametric cost determina-tion methods presents the best prerequisite for further development of the methodology. Furthermore, the conclusive aims of this paper are derived basically from the results of the study. The process of parametric cost modelling is described in detail with the aid of a systematic four-stage procedure. Initially, a structuring of the overall project is performed in individual cost segments. A hypothesis about the relevant model variables and their interactions is then established, a selection matrix with the essential product-, production- and project-specific influencing factors serving as an aid. Next, with the help of structure-testing statistical methods the hypothesis is transformed into an explanatory model. The information requirement that is connected with the statistical evaluation is limited to a minimum by grouping several influencing variables into repre-sentative multi-variables and by the targeted selection of statistical methods. The costs of future development projects can be forecasted on the basis of the explanatory model. The effects of uncertainty and stochastic influences on the forecast result are simulated with the aid of the Monte-Carlo-Method.